A method and system for inverting factors of fetal birth defects

By extracting and screening fetal birth defect characteristics, and using the principle of weight end elimination, the problem of inaccurate judgment of fetal birth defects in the prior art is solved, and accurate analysis of fetal birth defect types and causal identification are achieved.

CN118552796BActive Publication Date: 2024-11-26SHENZHEN UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411016863.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-26
Estimated Expiration
2044-07-29

AI Technical Summary

Technical Problem

The prior art cannot accurately extract data characteristics related to fetal birth defects, resulting in inaccurate judgment on attribution of the corresponding type of birth defects.

Method used

By obtaining maternal and fetal detection data, extracting and classifying fetal birth defect characteristics, screening the characteristics using the preset weight end elimination principle, obtaining the optimal subset of features, and calculating the inversion results of the factors that develop fetal birth defects based on this.

Benefits of technology

Accurate attribution of fetal birth defect types is achieved, the biggest influencing factors are found, the causal relationship between the occurrence and development of fetal birth defects and characteristics is analyzed, and the accuracy of judgment is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118552796B_ABST
    Figure CN118552796B_ABST
Patent Text Reader

Abstract

The present invention provides a method and system for inverting factors affecting the occurrence and development of fetal birth defects, which specifically relate to the technical field of medical data processing. The scheme includes: classifying fetal birth defect characteristics extracted from maternal and fetal detection data to obtain several fetal birth defect categories, and screening the characteristics in each fetal birth defect category according to a user-defined birth defect impact weight calculation and an impact weight bottom elimination principle to obtain an optimal feature subset. The present invention can accurately screen out key factors that affect the occurrence of a certain type of birth defect in the fetus according to a user-defined impact weight bottom elimination principle, thereby facilitating in-depth exploration of the corresponding fetal birth defect occurrence and development factor inversion results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method and system for inverting factors of occurrence and development of fetal birth defects. Background Art

[0002] At present, the research on fetal birth defects is mainly concentrated in clinical research, focusing on analyzing the health test data of a specific pregnant woman to obtain the key factors that affect the birth defects of the corresponding fetus, but it is impossible to accurately count the key factors that affect the prevalence of a certain type of birth defects in the fetus. Although some scholars use deep learning networks to conduct risk analysis on fetal birth defects, due to improper data feature extraction methods, it is impossible to accurately extract data features related to the corresponding type of fetal birth defects, which easily leads to inaccurate attribution judgments on the causes of the corresponding type of fetal birth defects. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a method and system for inverting the factors causing the occurrence and development of fetal birth defects, aiming to solve the problem in the prior art that the data features related to the corresponding type of fetal birth defects cannot be accurately extracted, resulting in inaccurate attribution judgment of the causes of the corresponding type of fetal birth defects.

[0004] In order to achieve the above-mentioned object, the first aspect of the present invention provides a method for inverting factors of occurrence and development of fetal birth defects, comprising:

[0005] Acquiring maternal and fetal test data, and extracting fetal birth defect characteristics from the maternal and fetal test data;

[0006] Classifying the fetal birth defect characteristics to obtain several fetal birth defect categories;

[0007] Screening the features in each fetal birth defect category according to a preset weighted bottom elimination principle to obtain an optimal feature subset;

[0008] Based on the optimal feature subset, the inversion results of the factors causing and developing fetal birth defects are obtained.

[0009] Optionally, the maternal-fetal detection data includes maternal-fetal text health data and maternal-fetal image health data, and the extracting of fetal birth defect characteristics from the maternal-fetal detection data includes:

[0010] Extracting defect features from the maternal-fetal text health data and the maternal-fetal image health data to obtain text defect features and image defect features respectively;

[0011] The text defect feature and the image defect feature are fused to obtain fetal birth defect features.

[0012] Optionally, the fetal birth defect characteristics are classified to obtain several fetal birth defect categories, including:

[0013] Using a multi-layer perceptron to classify the fetal birth defect characteristics, and obtain a probability distribution of the fetal birth defect categories;

[0014] Based on the probability distribution of the fetal birth defect categories, several fetal birth defect categories are obtained.

[0015] Optionally, the features in each fetal birth defect category are screened according to a preset weighted bottom elimination principle to obtain an optimal feature subset, including:

[0016] Calculating and sorting the impact weights of the features in each of the fetal birth defect categories to obtain an impact weight sequence corresponding to each of the fetal birth defect categories;

[0017] According to the preset weight end elimination principle, the features with less than a preset weight threshold in each of the influence weight sequences are eliminated to obtain a reduced subset corresponding to each of the influence weight sequences;

[0018] The performance of each of the reduced subsets is evaluated using a preset performance indicator to obtain an optimal feature subset.

[0019] Optionally, the calculating and sorting the influence weights of the features in each of the fetal birth defect categories to obtain an influence weight sequence corresponding to each of the fetal birth defect categories includes:

[0020] Using a recursive feature elimination algorithm, the feature with the lowest influence weight in each of the fetal birth defect categories is eliminated to obtain a reserved feature subset in each of the fetal birth defect categories;

[0021] Repeat the step of using a recursive feature elimination algorithm to eliminate the features with the lowest influence weight until the influence weight ranking of the features in all the fetal birth defect categories is completed, and an influence weight sequence corresponding to each fetal birth defect category is obtained.

[0022] Optionally, obtaining the inversion results of the factors causing and developing fetal birth defects based on the optimal feature subset includes:

[0023] Based on the optimal feature subset, obtaining several key influencing factors;

[0024] Calculate the influence weights of all the key influencing factors and sort them to obtain a key influencing factor sequence;

[0025] Based on the sequence of key influencing factors, the inversion results of factors affecting the occurrence and development of fetal birth defects are obtained.

[0026] A second aspect of the present invention provides a system for inverting factors of fetal birth defects, the system comprising:

[0027] An information acquisition module, used to acquire maternal and fetal test data and extract fetal birth defect characteristics from the maternal and fetal test data;

[0028] A classification module is used to classify the fetal birth defect characteristics to obtain a number of fetal birth defect categories;

[0029] An optimal feature screening module, used to screen the features in each fetal birth defect category according to a preset weighted bottom elimination principle to obtain an optimal feature subset;

[0030] The inversion result module is used to obtain the inversion results of the factors causing and developing fetal birth defects based on the optimal feature subset.

[0031] Optionally, the optimal feature screening module includes an influence weight ranking unit, an influence factor screening unit and a key feature evaluation unit, wherein:

[0032] An impact weight sorting unit, used to calculate and sort the impact weights of the features in each of the fetal birth defect categories to obtain an impact weight sequence corresponding to each of the fetal birth defect categories;

[0033] An influencing factor screening unit, used to eliminate the features with a weight lower than a preset weight threshold in each of the influencing weight sequences according to a preset weight end elimination principle, and obtain a reduced subset corresponding to each of the influencing weight sequences;

[0034] The key feature evaluation unit is used to evaluate the performance of each of the reduced subsets using a preset performance indicator to obtain an optimal feature subset.

[0035] The third aspect of the present invention provides an intelligent terminal, which includes a memory, a processor, and a program for inverting factors of the occurrence and development of fetal birth defects stored in the memory and executable on the processor, wherein the program for inverting factors of the occurrence and development of fetal birth defects, when executed by the processor, implements any step of the above-mentioned method for inverting factors of the occurrence and development of fetal birth defects.

[0036] A fourth aspect of the present invention provides a computer-readable storage medium, on which is stored a program for inverting factors affecting the occurrence and development of fetal birth defects. When the program for inverting factors affecting the occurrence and development of fetal birth defects is executed by a processor, the program implements any step of the above-mentioned method for inverting factors affecting the occurrence and development of fetal birth defects.

[0037] Compared with the prior art, the beneficial effects of this solution are as follows:

[0038] The present invention classifies fetal birth defect features extracted from maternal and fetal detection data to obtain several fetal birth defect categories, and screens the features in each fetal birth defect category according to a user-defined birth defect impact weight calculation and weight end elimination principle to obtain an optimal feature subset, thereby finding the greatest influencing factors of this type of birth defect, which is conducive to analyzing the causal relationship between the occurrence, development and outcome of these features and fetal birth defects, and obtaining the inversion results of the occurrence and development factors of fetal birth defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0040] Figure 1 This is a flow chart of the method for inverting the factors causing and developing fetal birth defects of the present invention;

[0041] Figure 2 Schematic diagram of the structure of the fusion attention module of the present invention;

[0042] Figure 3 A schematic diagram of the birth defect impact weight calculation and weight-end elimination mechanism of the present invention;

[0043] Figure 4 It is a schematic diagram of the birth defect impact weight calculation and weighted bottom elimination process of the present invention;

[0044] Figure 5 It is a schematic diagram of the module of the inversion system for factors causing and developing fetal birth defects of the present invention;

[0045] Figure 6 It is a schematic diagram of the structure of the intelligent terminal of the present invention. DETAILED DESCRIPTION

[0046] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0047] It should be understood that when used in this specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0048] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0049] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0050] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0052] In view of the existing problem that the data features related to the corresponding types of fetal birth defects cannot be accurately extracted, resulting in inaccurate attribution judgments on the causes of the corresponding types of fetal birth defects, the present invention proposes a method for inverting the occurrence and development factors of fetal birth defects. The method mainly includes: in the large model learning of maternal and fetal detection data and maternal and fetal characteristic data, the birth defect feature perception obtained is sent to a birth defect classifier for birth defect classification through the feature perception and fusion of maternal and fetal detection data and image data, and the influence weight sequence of each type of maternal and fetal queue data is obtained. The "weight tail elimination" principle is adopted to remove one-dimensional maternal and fetal data, and then the maternal and fetal birth defect classification weight is calculated. This cycle is repeated and finally recursively returned to some maternal and fetal health data with similar weights, that is, the maximum influencing factor of this type of birth defect is found, which is conducive to analyzing the causal relationship between the occurrence, development and outcome of these features and fetal birth defects, and obtaining the inversion results of the occurrence and development factors of fetal birth defects.

[0053] The embodiment of the present invention provides a method for inverting the factors affecting the occurrence and development of fetal birth defects, which is deployed on electronic devices such as computers and servers. The application scenario is to explore the key factors that affect the occurrence and development of fetal birth defects, and there is no restriction on the type of fetal birth defects. Specifically, Figure 1 As shown, the steps of the method in this embodiment include:

[0054] Step S100: acquiring maternal and fetal test data, and extracting fetal birth defect characteristics from the maternal and fetal test data;

[0055] Specifically, since fetal birth defects are closely related to the health status of pregnant women and fetal characteristic data in maternal and fetal detection data, this embodiment uses maternal and fetal detection data as the data source. Since the effective data features contained in the maternal and fetal detection data used to explore the key factors affecting the occurrence and development of fetal birth defects are as comprehensive as possible, this embodiment collects maternal and fetal detection data of pregnant women at various stages of the entire pregnancy cycle as the data source, and the maternal and fetal detection data can be maternal and fetal image health data expressed in unstructured forms such as maternal and fetal CT and B-ultrasound, or maternal and fetal text health data expressed in two-dimensional structured forms such as digital, character, and logical forms of maternal and fetal clinical detection, or a combination of different types of data. For multi-modal maternal and fetal detection data, it is necessary to extract corresponding features according to the feature extraction method of each modality of data, and then fuse the features of multiple modalities to obtain fetal birth defect features, which lays a good foundation for the subsequent use of fetal birth defect features to determine the type of birth defects and the size of the correlation. Furthermore, before fusing the features of each modality, preprocessing such as normalization or standardization is performed to ensure that the features of each modality are converted to the same scale, and then the channel dimension of the feature vector is used as the splicing dimension to fuse the feature vectors of different modalities into a high-dimensional joint feature vector to obtain the fetal birth defect feature. In order to avoid the feature dimension after splicing being too high, dimensionality reduction technology can also be used for dimensionality reduction processing to retain key information while reducing the amount of calculation.

[0056] In one embodiment, if the maternal-fetal detection data includes maternal-fetal text health data and maternal-fetal image health data, then the extraction of fetal birth defect features in the maternal-fetal detection data includes: extracting defect features from the maternal-fetal text health data and the maternal-fetal image health data to obtain text defect features and image defect features respectively; then concatenating the text defect features and the image defect features through a birth defect perception fusion network to form a joint feature vector, which contains information from both text and image modalities and can provide richer contextual information. Then, by Figure 2 The fusion attention module shown in the figure performs deep fusion on the joint feature vector. Specifically, the fusion attention module first converts the joint feature vector into a joint feature sequence and then decomposes it by parity. Downsampling is performed to separate the even elements from the odd elements. The two subsequences formed have a coarse time resolution but retain most of the information of the original time series. The process can be expressed as:

[0057] (1);

[0058] in, represents the joint feature sequence, represents the sequence length, Indicates the number of channel features. Represents the odd and even subsequences obtained after decomposition. However, the downsampling process may lead to potential information loss. To compensate for this potential problem, this module introduces an interactive learning strategy that allows information exchange between the two subsequences by learning the affine transformation parameters of the two subsequences. It mainly uses different convolution modules to project the two subsequences into different latent spaces and convert them into The scaling factor is obtained by mutual learning of neural networks and is used to amplify positive values ​​and reduce negative values ​​to achieve data enhancement. The process of interactive learning strategy can be expressed as:

[0059] (2);

[0060] in, represents the convolutional layer, To convert to exponential form, It is element-wise multiplication.

[0061] Then, the average pooling layer and the maximum pooling layer are used to compress the time dimension of the odd and even subsequences. The input matrix is ​​compressed to The sequence output of , while keeping the number of channel features unchanged, obtains a global feature and a local feature. Then the global feature and the local feature are sent to a multilayer perceptron (MLP) with shared weights for learning, and two The activation function used by MLP is the ReLU function. The number of layers of MLP is 2, and each layer outputs a feature sequence. The fusion is achieved by adding two feature sequences, that is:

[0062] (3);

[0063] in, is the feature sequence after weighted fusion, is the average pooling layer, is the maximum pooling layer. The average pooling and maximum pooling strategies are used to aggregate and strengthen feature information, where average pooling provides a global perspective on the features, while maximum pooling highlights the most critical local features. Specifically, the average pooling operation calculates the average value of each small area in the feature map to generate a new, smaller feature map, which tends to extract smoother and more general features; while the maximum pooling selects the maximum value of each small area in the feature map as the representative of the area, which tends to extract more specific and prominent features. By fusing the features obtained by these two pooling strategies, the network can integrate global contextual information and key local information, which can significantly improve the expressiveness of features and the recognition performance of the model. Through MLP learning the features of the channel dimension and the importance of each channel feature, deep abstraction and synthesis are performed, so that the model has the ability to learn complex patterns and relationships in the input features and form effective feature representations.

[0064] Furthermore, the birth defect perception fusion network adopts the attention weight mechanism to weight the fused feature map, and uses its powerful local pattern recognition and feature extraction capabilities to learn the channel information of the original sequence. One-dimensional convolution is used to accurately obtain the attention weight of each channel, and realize the effective interaction of multi-channel information to adapt to time series data with different numbers of channels. Specifically, a 1X1 convolution kernel is first used to extract features with unique temporal relationships from the subsequence to enhance the representation ability of the model, and then the nonlinear activation function ReLU function is used to learn complex feature information, and then a 1X1 convolution kernel is used to realize information interaction between channels. Finally, the sigmoid activation function is used to set the channel weight value between 0-1 to obtain the attention weight matrix. The specific process can be expressed as:

[0065] (4);

[0066] in, is the attention weight matrix, Represents the ReLU function activation function, is the Sigmoid activation function. The weights of different features reflect the importance of each feature in the current task, and the weighted feature map can be used for subsequent classification or other tasks. The attention weight mechanism enables the network to automatically learn and emphasize feature channels that are more useful for the current task, while suppressing less useful features, thereby improving the performance and generalization ability of the model.

[0067] Finally, the attention weight matrix is ​​multiplied by the channel of the original joint feature sequence to complete the channel enhancement. The weighting uses the element-by-element multiplication method. Each value in the attention weight matrix corresponds to the eigenvalue of the original data. By multiplying the weight value with the corresponding eigenvalue, the eigenvalue is weighted and adjusted to obtain the weighted fetal birth defect feature. ,Right now:

[0068] (5);

[0069] in, Represents element-by-element multiplication (Hadamard product). The weighted fetal birth defect features can be used for subsequent classification or other tasks. This weighting mechanism can effectively improve the recognition performance and accuracy of the model on specific tasks.

[0070] Step S200: classifying the fetal birth defect characteristics to obtain several fetal birth defect categories;

[0071] Specifically, the fetal birth defect characteristics are classified using a multi-layer perceptron to obtain a probability distribution of fetal birth defect categories; based on the probability distribution of the fetal birth defect categories, several fetal birth defect categories are obtained.

[0072] For the deeply fused fetal birth defect features output by the birth defect fusion network, the fully connected layer (FC) is used as a classifier to integrate the extracted fusion features into the final representation, and its output result is usually consistent with the number of classification categories. Then the output of the fully connected layer is transformed by the softmax activation function to obtain the probability distribution of each category. In order to solve the multi-classification problem, this embodiment trains a classifier for each category, and each classifier is a binary classification, which is responsible for distinguishing its corresponding category from other categories. For the same fetal birth defect feature data, all classifiers will give a score and convert it into a probability through the softmax function, which is equivalent to outputting a probability value for each category. By comparing the probability values ​​of these scores, the category corresponding to the highest probability is the prediction result of the model, thereby obtaining one or several fetal birth defect categories, which is conducive to the subsequent analysis of the key factors causing this type of fetal birth defect for the maternal and fetal cohort data corresponding to each fetal birth defect category.

[0073] Step S300: Screening the features in each fetal birth defect category according to a preset weighted bottom elimination principle to obtain an optimal feature subset;

[0074] Specifically, the influence weights of the features in each fetal birth defect category are calculated and sorted to obtain an influence weight sequence corresponding to each fetal birth defect category; according to the preset weight end elimination principle, the features in each influence weight sequence that are less than a preset influence weight threshold are eliminated to obtain a reduced subset corresponding to each influence weight sequence; the performance of each reduced subset is evaluated using preset performance indicators to obtain the optimal feature subset.

[0075] The birth defect impact weight calculation and "weighted tail elimination" mechanism adopted in this embodiment are as follows: Figure 3 As shown. In the calculation of the impact weights for evaluating fetal birth defects, the recursive feature elimination method (RFE) is used as the core evaluator. The recursive feature elimination method is an optimization algorithm based on a greedy strategy, and its goal is to identify and select the feature subset that contributes most to the model performance. The algorithm repeatedly creates models, retains the best features in each iteration, and removes the features with the lowest impact weights in each fetal birth defect category to obtain the reserved feature subset in each fetal birth defect category; in the next iteration, the features that were not selected in the previous modeling are used to build the next model, and so on, repeating the steps of using the recursive feature elimination algorithm to remove the features with the lowest impact weights until the impact weights of all features in the fetal birth defect category are ranked, and the robustness of the evaluation is ensured by cross-validation, so as to select the optimal number of important features.

[0076] The recursive feature elimination algorithm uses the "weighted tail elimination" mechanism to select the most predictive features from all possible input features, such as Figure 4 As shown, the details are as follows:

[0077] In the recursive feature elimination stage, all fetal birth defect features are first eliminated. As a candidate feature subset, Indicates fetal birth defect characteristics, . Using the birth defect risk classification network as the base model, the model is trained with a subset of candidate features to determine the association weights between each feature and birth defects. During the model training process, each feature is assigned an importance score to reflect the relative importance of the corresponding feature in the classification of birth defects. After the training is completed, all features will be ranked by importance score. Under the framework of the recursive feature elimination algorithm, the top-ranked features are considered to contribute the most to the model's predictive ability, while the bottom-ranked features are considered to be the least important. Next, the feature with the lowest importance score is removed from the current feature set. , , forming a new reduced feature subset The recursive training process is repeated until the importance rating of all features is completed, so that different numbers of features are selected according to the importance of the features to form multiple important feature subsets. It can be seen that the order in which the features are removed reflects the importance of each feature on the model performance, and the features removed first are considered to be the least important features.

[0078] In the cross-validation stage, the performance indicators of each important feature subset (such as accuracy, recall rate, etc.) are compared, all subsets are sorted according to performance, and the important feature subset with the best performance, that is, the optimal feature subset, is selected. This subset contains a batch of maternal and fetal feature data with the greatest influence, which is the key factor affecting this type of fetal birth defect.

[0079] Step S400: Based on the optimal feature subset, the inversion results of the factors causing and developing fetal birth defects are obtained.

[0080] Specifically, based on the optimal feature subset, several key influencing factors are obtained; the influence weights of all the key influencing factors are calculated and sorted to obtain a key influencing factor sequence, that is, a size-ranked list of key factors affecting fetal birth defects can be obtained based on the importance score of each feature in the feature subset with the best performance that is finally selected; and some features with relatively large contributions are determined based on the sorting, which is conducive to analyzing the causal relationship between these features and the occurrence, development and outcome of the corresponding types of fetal birth defects, and obtaining the inversion results of the factors causing the occurrence and development of fetal birth defects.

[0081] like Figure 5 As shown, corresponding to the above-mentioned method for inverting the occurrence and development factors of fetal birth defects, an embodiment of the present invention further provides a system for inverting the occurrence and development factors of fetal birth defects, and the above-mentioned system for inverting the occurrence and development factors of fetal birth defects includes:

[0082] The information acquisition module 510 is used to acquire maternal and fetal test data and extract fetal birth defect characteristics from the maternal and fetal test data;

[0083] A classification module 520 is used to classify the fetal birth defect characteristics to obtain a number of fetal birth defect categories;

[0084] The optimal feature screening module 530 is used to screen the features in each fetal birth defect category according to a preset weighted bottom elimination principle to obtain an optimal feature subset;

[0085] The inversion result module 540 is used to obtain the inversion results of the factors causing and developing fetal birth defects based on the optimal feature subset.

[0086] Furthermore, the optimal feature screening module 530 includes an influence weight sorting unit, an influence factor screening unit and a key feature evaluation unit, wherein:

[0087] An impact weight sorting unit, used to calculate and sort the impact weights of the features in each of the fetal birth defect categories to obtain an impact weight sequence corresponding to each of the fetal birth defect categories;

[0088] An influencing factor screening unit, used to eliminate the features with a weight lower than a preset weight threshold in each of the influencing weight sequences according to a preset weight end elimination principle, and obtain a reduced subset corresponding to each of the influencing weight sequences;

[0089] The key feature evaluation unit is used to evaluate the performance of each of the reduced subsets using a preset performance indicator to obtain an optimal feature subset.

[0090] Specifically, in this embodiment, the specific functions of the above-mentioned fetal birth defect occurrence and development factor inversion system can also refer to the corresponding description in the above-mentioned fetal birth defect occurrence and development factor inversion method, which will not be repeated here.

[0091] Based on the above embodiments, the present invention further provides an intelligent terminal, whose principle block diagram can be shown as follows: Figure 6 As shown. The above-mentioned intelligent terminal includes a processor, a memory, a network interface and a display screen connected through a system bus. Among them, the processor of the intelligent terminal is used to provide computing and control capabilities. The memory of the intelligent terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and an inversion program for the occurrence and development of fetal birth defects. The internal memory provides an environment for the operation of the operating system in the non-volatile storage medium and the inversion program based on the occurrence and development of fetal birth defects. The network interface of the intelligent terminal is used to communicate with an external terminal through a network connection. When the inversion program for the occurrence and development of fetal birth defects is executed by the processor, the steps of any one of the above-mentioned methods for inverting the occurrence and development of fetal birth defects are implemented. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen.

[0092] Those skilled in the art will understand that Figure 6 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the smart terminal to which the scheme of the present invention is applied. The specific smart terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0093] In one embodiment, a smart terminal is provided, which includes a memory, a processor, and a program for inverting factors of the occurrence and development of fetal birth defects stored in the memory and executable on the processor. When the program for inverting factors of the occurrence and development of fetal birth defects is executed by the processor, the steps of any one of the methods for inverting factors of the occurrence and development of fetal birth defects provided in the embodiments of the present invention are implemented.

[0094] An embodiment of the present invention also provides a computer-readable storage medium, on which is stored a program for inverting factors for the occurrence and development of fetal birth defects. When the program for inverting factors for the occurrence and development of fetal birth defects is executed by a processor, the steps of any one of the methods for inverting factors for the occurrence and development of fetal birth defects provided in an embodiment of the present invention are implemented.

[0095] It should be understood that the serial numbers of the steps in the above embodiments do not imply a sequence of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0096] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the above-mentioned device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0097] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0098] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0099] In the embodiments provided by the present invention, it should be understood that the disclosed apparatus / terminal equipment and method can be implemented in other ways. For example, the apparatus / terminal equipment embodiments described above are only illustrative, for example, the division of the above modules or units is only a logical function division, and in actual implementation, other division methods can be used, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0100] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features therein may be replaced by equivalents. However, these modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for inverting factors of fetal birth defects, characterized by: The following steps are involved: Acquiring maternal and fetal test data, and extracting fetal birth defect characteristics from the maternal and fetal test data; Classifying the fetal birth defect characteristics to obtain several fetal birth defect categories; Screening the features in each fetal birth defect category according to a preset weighted elimination principle to obtain an optimal feature subset; Based on the optimal feature subset, obtaining an inversion result of factors causing and developing fetal birth defects; The maternal-fetal detection data includes maternal-fetal text health data and maternal-fetal image health data, and extracting fetal birth defect characteristics from the maternal-fetal detection data includes: Extract defect features from the maternal-fetal text health data and the maternal-fetal image health data to obtain text defect features and image defect features respectively; splice and transform the text defect features and the image defect features to obtain a joint feature sequence; downsample the joint feature sequence through parity decomposition to generate odd subsequences and even subsequences; use different convolution modules to project the odd subsequences and even subsequences into different latent spaces, and convert them into exponential forms for scaling transformation, and use average pooling layers and maximum pooling layers to compress the time dimensions of the odd subsequences and even subsequences to obtain a global feature and a local feature while keeping the number of channel features unchanged; use a multi-layer perceptron with shared weights to perform weighted fusion of the global features and the local features, and use one-dimensional convolution to obtain the attention weight of each channel to obtain an attention weight matrix; multiply the attention weight matrix by the channel of the joint feature sequence to obtain fetal birth defect features; The fetal birth defect characteristics are classified to obtain several fetal birth defect categories, including: A classifier is trained for each fetal birth defect category, and each classifier is a binary classification. For the same fetal birth defect feature data, all classifiers output a score, which is converted into a probability. By comparing the probabilities corresponding to the scores output by the classifiers, the category corresponding to the highest probability is selected as the fetal birth defect category. The features in each fetal birth defect category are screened according to a preset weighted elimination principle to obtain an optimal feature subset, including: Calculating and sorting the impact weights of the features in each fetal birth defect category to obtain an impact weight sequence corresponding to each fetal birth defect category; eliminating features with less than a preset impact weight threshold in each impact weight sequence according to a preset weight end elimination principle to obtain a reduced subset corresponding to each impact weight sequence; and evaluating the performance of each of the reduced subsets using a preset performance indicator to obtain an optimal feature subset; The calculating and sorting the influence weights of the features in each fetal birth defect category to obtain an influence weight sequence corresponding to each fetal birth defect category includes: Using a recursive feature elimination algorithm, the features with the lowest influence weight in each of the fetal birth defect categories are eliminated to obtain a reserved feature subset in each of the fetal birth defect categories; repeatedly performing the step of eliminating the features with the lowest influence weight using the recursive feature elimination algorithm until the influence weight ranking of the features in all of the fetal birth defect categories is completed to obtain an influence weight sequence corresponding to each of the fetal birth defect categories.

2. The method for inverting the factors causing and developing fetal birth defects according to claim 1, characterized in that: The step of obtaining an inversion result of factors causing and developing fetal birth defects based on the optimal feature subset includes: Based on the optimal feature subset, several key influencing factors are obtained; Calculate the influence weights of all the key influencing factors and sort them to obtain a key influencing factor sequence; Based on the key influencing factor sequence, the inversion results of the factors causing and developing fetal birth defects are obtained.

3. A fetal birth defect occurrence and development factor inversion system, characterized by: The method for inverting the factors causing and developing fetal birth defects according to any one of claims 1 to 2 is applied to the system, comprising: An information acquisition module, configured to acquire maternal and fetal test data and extract fetal birth defect characteristics from the maternal and fetal test data; A classification module is used to classify the fetal birth defect characteristics to obtain a number of fetal birth defect categories; An optimal feature screening module is used to screen the features in each fetal birth defect category according to a preset weighted bottom elimination principle to obtain an optimal feature subset; The inversion result module is used to obtain the inversion results of the factors causing and developing fetal birth defects based on the optimal feature subset.

4. An intelligent terminal, characterized in that: The intelligent terminal includes a memory, a processor, and a program for inverting the factors causing and developing fetal birth defects, which is stored in the memory and can be run on the processor. When the program for inverting the factors causing and developing fetal birth defects is executed by the processor, the steps of the method for inverting the factors causing and developing fetal birth defects as described in any one of claims 1-2 are implemented.

5. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program for inverting factors of the occurrence and development of fetal birth defects. When the program for inverting factors of the occurrence and development of fetal birth defects is executed by a processor, the steps of the method for inverting factors of the occurrence and development of fetal birth defects as described in any one of claims 1-2 are implemented.