Prenatal diagnosis genetic nursing system based on big data
By expanding data and optimizing the community in the construction of genetic care communities, the problems of insufficient data diversity, inaccurate community division and low search efficiency in traditional methods are solved, and more efficient and accurate construction and optimization of genetic care communities are achieved.
Patent Information
- Application Number
- CN202510194211.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-21
- Publication Date
- 2025-03-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The traditional genetic care community construction method has problems such as insufficient data diversity, low extension data quality, inaccurate community division, lack of dynamic adjustment mechanism, low search efficiency and local optimal solutions.
Data expansion is performed through data division, designing data expansion model architecture, designing adversarial loss function, designing cyclic loss function, designing feature loss function and model comprehensive training; defining prenatal diagnostic feature similarity, basic gravitational function, initial gravitational graph, community center and community adjustment factors to generate genetic care communities; using search initialization, community quality assessment function, dynamic guidance factor, dynamic evolutionary search and reset strategies to optimize community parameters.
It improves data diversity and scalability, enhances the accuracy and flexibility of community division, improves search efficiency and global search capabilities, and ensures the stability and reliability of optimized results.
Smart Images
Figure CN119670809A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of genetic care technology, and in particular to a prenatal diagnosis genetic care system based on big data. Background Art
[0002] The prenatal diagnosis genetic care system based on big data uses big data technology and artificial intelligence technology to build a genetic care community model based on historical maternal prenatal diagnosis data and care plan data, and combines the community model to provide targeted genetic care guidance plans for pregnant women's families. However, the traditional genetic care community construction method has problems such as insufficient data diversity, low quality of extended data, and difficulty in maintaining consistency and feature retention between data; the traditional genetic care community construction method has problems such as inaccurate community division, lack of dynamic adjustment mechanism, and difficulty in effectively handling complex feature similarities; the traditional community optimization method has problems such as low search efficiency, easy to fall into local optimal solutions, lack of dynamic adjustment capabilities, and poor adaptability to complex community structures. Summary of the invention
[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a prenatal diagnosis genetic care system based on big data. In view of the problems of insufficient data diversity, low quality of expanded data, and difficulty in maintaining consistency and feature retention in traditional genetic care community construction methods, this method expands data through data partitioning, designing data expansion model architecture, designing adversarial loss function, designing cyclic loss function, designing feature loss function and comprehensive model training, thereby improving data diversity, effect and quality of data expansion; in view of the problems of inaccurate community partitioning, lack of dynamic adjustment mechanism and difficulty in effectively handling complex feature similarity in traditional genetic care community construction methods, this scheme defines prenatal diagnosis feature similarity, defines basic gravity function The genetic nursing community is generated by counting, creating an initial gravity graph, connecting data, setting the initial community center, calculating the correlation, allocating data points, designing community adjustment factors and updating the community center, which improves the accuracy, stability, flexibility and adaptability of community division; in view of the problems of low search efficiency, easy to fall into local optimal solutions, lack of dynamic adjustment capabilities and poor adaptability to complex community structures in traditional community optimization methods, this scheme searches for high-quality community parameters through search initialization, defining community quality evaluation functions, designing dynamic guiding factors, dynamic evolutionary search, designing reset strategies and designing search rules, avoiding blind search, improving search efficiency, enhancing global search capabilities, enhancing the flexibility and adaptability of the search process, and ensuring the stability and reliability of the optimization results.
[0004] The technical solution adopted by the present invention is as follows: the prenatal diagnosis genetic care system based on big data provided by the present invention includes a data acquisition module, a data expansion module, a genetic care community construction module, a community optimization module, and a genetic care plan generation module;
[0005] The data collection module collects the prenatal diagnosis data and nursing plan data of the pregnant woman, wherein the prenatal diagnosis data includes serological test data, genetic test data, ultrasound examination data and family genetic disease history data; the nursing plan data includes life care measures, drug management measures, psychological care measures and education measures.
[0006] Furthermore, the data expansion module performs data expansion by data partitioning, designing a data expansion model architecture, designing an adversarial loss function, designing a cyclic loss function, designing a feature loss function, and model comprehensive training, specifically including the following contents:
[0007] Data partitioning: creating two data domains: one is the source domain containing all original data, and the other is the extended domain for storing all expanded data.
[0008] Design the data expansion model architecture, use the long short-term memory network as the basic architecture, and build two generators and one discriminator; the first generator includes an input layer, an LSTM layer, a fully connected layer, and an output layer; the LSTM layer is activated by a hyperbolic tangent function, and the fully connected layer is activated by a linear rectifier function; the second generator has the same structure as the first generator but in the opposite direction; the fully connected layer of the first discriminator is activated by a sigmoid function, and the rest of the structure is the same as the first generator;
[0009] Design the adversarial loss function, which is expressed as follows:
[0010] ;
[0011] in, represents a first generator that converts source domain data into extended domain data, represents the first discriminator function, represents the adversarial loss function, Indicates extended domain data. Indicates the extended domain data distribution, represents the first discriminator function, represents the discrimination result of the first discriminator on the extended domain data, represents the logarithmic function, represents the expected output of the logarithm of the first discriminator’s discriminant result on the extended domain data, Represents source domain data, represents the source domain data distribution, represents the conditional vector, represents the data generated by combining the source domain data and the conditional vector, Represents the data generated by the first discriminator The expected output of the judgment result is to take the opposite number, add one, and then take the logarithm;
[0012] Design the cycle loss function, which is expressed as follows:
[0013] ;
[0014] in, represents a second generator that converts the extended domain data into the source domain data, represents the cycle loss function, Represents source domain data After being converted to the extended domain by the first generator, the first reconstruction error is converted back to the source domain by the second generator. Indicates extended domain data After being converted to the source domain by the second generator, the second reconstruction error is converted back to the extended domain by the first generator;
[0015] Design feature loss function, expressed as follows:
[0016]
[0017] in, represents the feature loss function, Indicates extended domain data Input to the first generator When the error between the output data and the original data is Represents source domain data Input to the second generator When , the error between the output data and the original data;
[0018] Model comprehensive training, uses three loss functions to conduct comprehensive training of the data expansion model, which is expressed as follows:
[0019] ;
[0020] in, The comprehensive loss value of the model, and Indicates importance weight;
[0021] Data expansion: Use the trained generator to generate prenatal diagnosis data and care plan data for pregnant women to expand the original dataset.
[0022] Furthermore, the genetic care community construction module generates a genetic care community by defining the similarity of prenatal diagnosis features, defining a basic gravity function, creating an initial gravity graph, data connection, setting an initial community center, calculating association, data point allocation, designing a community adjustment factor, and updating a community center, specifically including the following:
[0023] The similarity of prenatal diagnosis features is defined as follows:
[0024] ;
[0025] Among them, i and j represent the index of the data sample, and represent the i-th and j-th data samples respectively, represents the similarity of prenatal diagnosis features between the i-th and j-th data samples, f represents the index of prenatal diagnosis features, represents the total number of prenatal diagnostic features, represents the similarity weight corresponding to the f-th feature, represents the fth eigenvalue of the i-th data, represents the fth eigenvalue of the jth data, Indicates square root;
[0026] Define the basic gravitational function as follows:
[0027] ;
[0028] in, represents the basic gravitational strength between the i-th and j-th data samples, e represents a natural constant, and represents the gravitational transformation factor;
[0029] Create an initial gravity map, create an empty map , where D is the set of data samples and Q is the set of edges;
[0030] Data connection, calculate the gravitational force between each pair of data samples, in the figure Add an edge connecting the data pair and set the gravity value as the weight of the edge;
[0031] Set the initial community center and select data points as the initial community center;
[0032] Calculate the relevance and calculate the relevance of each data point to each community center, expressed as follows:
[0033] ;
[0034] in, and Respectively represent the data points and community centers for calculating the association degree, Represents data points With community center The correlation between represents the weight of the edge between the data point and the community center, Represents data points The fth eigenvalue of Indicates community center The fth eigenvalue of Represents data points With community center Similarity of prenatal diagnosis characteristics between
[0035] Data point allocation, where the data point is allocated to the community to which the community center with the greatest correlation with the data point belongs;
[0036] Design the community adjustment factor, set the Euclidean distance between the two farthest data points as the community width, calculate the value obtained by dividing the total number of data points in the community by the community width, set this value as the density value of the community, and perform dynamic adjustment of the community, as shown below:
[0037] ;
[0038] in, represents the community adjustment factor, and represents the scale adjustment factor, Represents the total number of data points in the community, Indicates the number of data points whose distance from the community center is less than 0.3 times the community width. Indicates the current density, and denote the minimum and maximum community density, respectively;
[0039] Community center update: calculate the mean of the positions of all data points in a community to obtain the average position of the community, fine-tune the average position, select the data point closest to the fine-tuned position, and set it as the latest community center, as shown below:
[0040] ;
[0041] in, Indicates the new community center, It means taking the nearest data point. represents the average location of the community, represents the community update weight, represents the symbolic function, represents the fine-tuning threshold, Indicates the old community center, Indicates taking the absolute value;
[0042] Generate communities, iteratively assign data points and update community centers, and assign data points to communities until the communities converge and form a stable community structure.
[0043] Furthermore, the community optimization module searches for high-quality community parameters by searching for initialization, defining community quality evaluation functions, designing dynamic guiding factors, dynamic evolutionary search, designing reset strategies, and designing search rules, thereby optimizing the community, which specifically includes the following contents:
[0044] Initialize the search, determine the search parameters, including the number of communities and the location of the initial community center, create a parameter search space, and generate an initial parameter search point cluster;
[0045] The community quality evaluation function is defined as follows:
[0046] ;
[0047] in, Indicates the quality of the community, and They represent the community similarity weight and discreteness weight respectively, m represents the index of the community, and n represents the index of the data point in the community. and Represent the total number of communities and the total number of data points in the community, respectively. Indicates the Euclidean distance. represents the nth data point in the mth community, represents the mth community center, represents the m+1th community center;
[0048] Design a dynamic guidance factor, expressed as follows:
[0049] ;
[0050] in, represents the dynamic guidance factor, t represents the current parameter search times, T represents the maximum parameter search times, , and represents the bootstrap constant, Represents the Euclidean distance between the current search point and the search point with the highest community quality;
[0051] Dynamic evolutionary search uses dynamic guiding factors to design parameter search strategies, which are expressed as follows:
[0052] ;
[0053] in, Indicates the position of the parameter point obtained by dynamic evolution search during the t+1th parameter search, Indicates the position of the parameter point obtained by dynamic evolution search during the tth parameter search, Represents a random number between 0 and 1. represents the average position of the search point cluster generated by the search initialization, is the position of the search point with the highest community quality among the searched parameter points, and check(·) means taking the nearest integer;
[0054] Design a reset strategy and adjust the search direction through the reset strategy, as shown below:
[0055] ;
[0056] in, Indicates the position of the parameter point obtained by executing the reset strategy during the t+1th parameter search. represents the position of the parameter point obtained by executing the reset strategy during the t-th optimization search. Represents a random number ranging from 0 to 1;
[0057] Design search rules, set the maximum number of parameter searches, set the community quality standard value, calculate the community quality corresponding to the initial parameter search point cluster, perform dynamic evolutionary search on the parameter search point cluster, calculate the community quality corresponding to the searched parameters, and if the community quality corresponding to more than 70% of the parameter points has been improved, perform the next dynamic evolutionary search, otherwise execute the reset strategy first, and then perform the next dynamic evolutionary search; if a parameter point with a community quality greater than the community quality standard value appears during the search process, output the parameters of the search point with the highest community quality among the parameter points; if the maximum number of parameter searches is reached, perform parameter search again; otherwise, continue searching.
[0058] Furthermore, the genetic care plan generation module inputs the new pregnant woman data into the constructed community model, reconstructs the genetic care community, obtains the community data to which the new pregnant woman data belongs, calculates the correlation between all data samples in the community and the new pregnant woman data, sets the care plan data of the data sample with the highest correlation as the care plan for the new pregnant woman, and provides genetic care for the pregnant woman.
[0059] The beneficial effects achieved by the present invention using the above scheme are as follows:
[0060] (1) In view of the problems of insufficient data diversity, low quality of expanded data, and difficulty in maintaining consistency and feature retention among data in the traditional genetic nursing community construction method, this paper expands data by partitioning data, designing data expansion model architecture, designing adversarial loss function, designing cyclic loss function, designing feature loss function and comprehensive model training, thereby improving data diversity, effect and quality of data expansion.
[0061] (2) In view of the problems of inaccurate community division, lack of dynamic adjustment mechanism and difficulty in effectively handling complex feature similarity in traditional genetic care community construction methods, this scheme generates genetic care communities by defining prenatal diagnosis feature similarity, defining basic gravity function, creating initial gravity graph, data connection, setting initial community center, calculating association, data point allocation, designing community adjustment factor and community center update, thereby improving the accuracy, stability, flexibility and adaptability of community division.
[0062] (3) In view of the problems of low search efficiency, easy to fall into local optimal solutions, lack of dynamic adjustment capabilities and poor adaptability to complex community structures in traditional community optimization methods, this scheme searches for high-quality community parameters through search initialization, definition of community quality evaluation function, design of dynamic guidance factors, dynamic evolutionary search, design of reset strategy and design of search rules, thus avoiding blind search, improving search efficiency, enhancing global search capabilities, enhancing the flexibility and adaptability of the search process, and ensuring the stability and reliability of the optimization results. BRIEF DESCRIPTION OF THE DRAWINGS
[0063] Figure 1 A schematic diagram of a prenatal diagnosis and genetic care system based on big data provided by the present invention;
[0064] Figure 2 This is a schematic diagram of the data expansion module;
[0065] Figure 3 Schematic diagram of building blocks for the genetics care community;
[0066] Figure 4 Schematic diagram of the module optimized for the community.
[0067] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION
[0068] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0069] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.
[0070] Example 1, see Figure 1 The prenatal diagnosis genetic care system based on big data provided by the present invention includes a data acquisition module, a data expansion module, a genetic care community construction module, a community optimization module, and a genetic care program generation module;
[0071] The data collection module collects the prenatal diagnosis data and nursing plan data of the pregnant woman;
[0072] The data expansion module performs data expansion by data partitioning, designing a data expansion model architecture, designing an adversarial loss function, designing a cyclic loss function, designing a feature loss function, and model comprehensive training;
[0073] The genetic care community building module generates a genetic care community by defining the similarity of prenatal diagnosis features, defining a basic gravity function, creating an initial gravity graph, data connection, setting an initial community center, calculating association, data point allocation, designing a community adjustment factor, and updating a community center;
[0074] The community optimization module searches for high-quality parameters for the community by searching for initialization, defining a community quality evaluation function, designing a dynamic guiding factor, dynamic evolutionary search, designing a reset strategy, and designing search rules;
[0075] The genetic care plan generation module inputs the new pregnant woman data into the constructed community model, calculates the correlation between all data samples in the community to which the new pregnant woman data belongs and the new pregnant woman data, sets the care plan data of the data sample with the highest correlation as the care plan for the new pregnant woman, and provides genetic care for the pregnant woman.
[0076] Example 2, see Figure 1This embodiment is based on the above embodiment. The data acquisition module collects the prenatal diagnosis data and nursing plan data of the pregnant woman. The prenatal diagnosis data includes serological test data, genetic test data, ultrasound examination data and family genetic disease history data; the nursing plan data includes life care measures, drug management measures, psychological care measures and education measures.
[0077] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment, and the data expansion module specifically includes the following contents:
[0078] Data partitioning: creating two data domains: one is the source domain containing all original data, and the other is the extended domain for storing all expanded data.
[0079] Design the data expansion model architecture, use the long short-term memory network as the basic architecture, and build two generators and one discriminator; the first generator includes an input layer, an LSTM layer, a fully connected layer, and an output layer; the LSTM layer is activated by a hyperbolic tangent function, and the fully connected layer is activated by a linear rectifier function; the second generator has the same structure as the first generator but in the opposite direction; the fully connected layer of the first discriminator is activated by a sigmoid function, and the rest of the structure is the same as the first generator;
[0080] Design the adversarial loss function, which is expressed as follows:
[0081] ;
[0082] in, represents a first generator that converts source domain data into extended domain data, represents the first discriminator function, represents the adversarial loss function, Indicates extended domain data. represents the extended domain data distribution, represents the first discriminator function, represents the discrimination result of the first discriminator on the extended domain data, represents the logarithmic function, represents the expected output of the logarithm of the first discriminator’s discriminant result on the extended domain data, Represents source domain data, represents the source domain data distribution, represents the conditional vector, represents the data generated by combining the source domain data and the conditional vector, Represents the data generated by the first discriminator The expected output of the judgment result is to take the opposite number, add one, and then take the logarithm;
[0083] Design the cycle loss function, which is expressed as follows:
[0084] ;
[0085] in, represents a second generator that converts the extended domain data into the source domain data, represents the cycle loss function, Represents source domain data After being converted to the extended domain by the first generator, the first reconstruction error is converted back to the source domain by the second generator. Indicates extended domain data After being converted to the source domain by the second generator, the second reconstruction error is converted back to the extended domain by the first generator;
[0086] Design feature loss function, expressed as follows:
[0087]
[0088] in, represents the feature loss function, Indicates extended domain data Input to the first generator When the error between the output data and the original data is Represents source domain data Input to the second generator When , the error between the output data and the original data;
[0089] Model comprehensive training, uses three loss functions to conduct comprehensive training of the data expansion model, which is expressed as follows:
[0090] ;
[0091] in, The comprehensive loss value of the model, and Indicates importance weight;
[0092] Data expansion: Use the trained generator to generate prenatal diagnosis data and care plan data for pregnant women to expand the original dataset.
[0093] By performing the above operations, in order to address the problems of insufficient data diversity, low quality of expanded data, and difficulty in maintaining consistency and feature retention among data in traditional genetic nursing community construction methods, this paper expands data through data partitioning, designing data expansion model architecture, designing adversarial loss function, designing cyclic loss function, designing feature loss function and comprehensive model training, thereby improving data diversity, effect and quality of data expansion.
[0094] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment, and the genetic care community building module specifically includes the following contents:
[0095] The similarity of prenatal diagnosis features is defined as follows:
[0096] ;
[0097] Among them, i and j represent the index of the data sample, and represent the i-th and j-th data samples respectively, represents the similarity of prenatal diagnosis features between the i-th and j-th data samples, f represents the index of prenatal diagnosis features, represents the total number of prenatal diagnostic features, represents the similarity weight corresponding to the f-th feature, represents the fth eigenvalue of the i-th data, represents the fth eigenvalue of the jth data, Indicates square root;
[0098] Define the basic gravitational function as follows:
[0099] ;
[0100] in, represents the basic gravitational strength between the i-th and j-th data samples, e represents a natural constant, and represents the gravitational transformation factor;
[0101] Create an initial gravity map, create an empty map , where D is the set of data samples and Q is the set of edges;
[0102] Data connection, calculate the gravitational force between each pair of data samples, in the figure Add an edge connecting the data pair and set the gravity value as the weight of the edge;
[0103] Set the initial community center and select data points as the initial community center;
[0104] Calculate the relevance and calculate the relevance of each data point to each community center, as shown below:
[0105] ;
[0106] in, and Respectively represent the data points and community centers for calculating the association degree, Represents data points With community center The correlation between represents the weight of the edge between the data point and the community center, Represents data points The fth eigenvalue of Indicates community center The fth eigenvalue of Represents data points With community center Similarity of prenatal diagnosis characteristics between
[0107] Data point allocation, where the data point is allocated to the community to which the community center with the greatest correlation with the data point belongs;
[0108] Design the community adjustment factor, set the Euclidean distance between the two farthest data points as the community width, calculate the value obtained by dividing the total number of data points in the community by the community width, set this value as the density value of the community, and perform dynamic adjustment of the community, as shown below:
[0109] ;
[0110] in, represents the community adjustment factor, and represents the scale adjustment factor, Represents the total number of data points in the community, Indicates the number of data points whose distance from the community center is less than 0.3 times the community width. Indicates the current density, and denote the minimum and maximum community density, respectively;
[0111] Community center update: calculate the mean of the positions of all data points in a community to obtain the average position of the community, fine-tune the average position, select the data point closest to the fine-tuned position, and set it as the latest community center, as shown below:
[0112] ;
[0113] in, Indicates the new community center, It means taking the nearest data point. represents the average location of the community, represents the community update weight, represents the symbolic function, represents the fine-tuning threshold, Indicates the old community center, Indicates taking the absolute value;
[0114] Generate communities, iteratively assign data points and update community centers, and assign data points to communities until the communities converge and form a stable community structure.
[0115] By performing the above operations, in order to address the problems of inaccurate community division, lack of dynamic adjustment mechanism and difficulty in effectively handling complex feature similarities in traditional genetic care community construction methods, this scheme generates genetic care communities by defining prenatal diagnosis feature similarity, defining basic gravity functions, creating initial gravity graphs, data connection, setting initial community centers, calculating associations, data point allocation, designing community adjustment factors and updating community centers, thereby improving the accuracy, stability, flexibility and adaptability of community division.
[0116] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment, and the community optimization module specifically includes the following contents:
[0117] Initialize the search, determine the search parameters, including the number of communities and the location of the initial community center, create a parameter search space, and generate an initial parameter search point cluster;
[0118] The community quality evaluation function is defined as follows:
[0119] ;
[0120] in, Indicates the quality of the community, and They represent the community similarity weight and discreteness weight respectively, m represents the index of the community, and n represents the index of the data point in the community. and Represent the total number of communities and the total number of data points in the community, respectively. Indicates the Euclidean distance. represents the nth data point in the mth community, represents the mth community center, represents the m+1th community center;
[0121] Design a dynamic guidance factor, expressed as follows:
[0122] ;
[0123] in, represents the dynamic guidance factor, t represents the current parameter search times, T represents the maximum parameter search times, , and represents the bootstrap constant, Represents the Euclidean distance between the current search point and the search point with the highest community quality;
[0124] Dynamic evolutionary search uses dynamic guiding factors to design parameter search strategies, which are expressed as follows:
[0125] ;
[0126] in, Indicates the position of the parameter point obtained by dynamic evolution search during the t+1th parameter search, Indicates the position of the parameter point obtained by dynamic evolution search during the tth parameter search, Represents a random number between 0 and 1. represents the average position of the search point cluster generated by the search initialization, is the position of the search point with the highest community quality among the searched parameter points, and check(·) means taking the nearest integer;
[0127] Design a reset strategy and adjust the search direction through the reset strategy, as shown below:
[0128] ;
[0129] in, Indicates the position of the parameter point obtained by executing the reset strategy during the t+1th parameter search. represents the position of the parameter point obtained by executing the reset strategy during the t-th optimization search. Represents a random number ranging from 0 to 1;
[0130] Design search rules, set the maximum number of parameter searches, set the community quality standard value, calculate the community quality corresponding to the initial parameter search point cluster, perform dynamic evolutionary search on the parameter search point cluster, calculate the community quality corresponding to the searched parameters, and if the community quality corresponding to more than 70% of the parameter points has been improved, perform the next dynamic evolutionary search, otherwise execute the reset strategy first, and then perform the next dynamic evolutionary search; if a parameter point with a community quality greater than the community quality standard value appears during the search process, output the parameters of the search point with the highest community quality among the parameter points; if the maximum number of parameter searches is reached, perform parameter search again; otherwise, continue searching.
[0131] By performing the above operations, in view of the problems of low search efficiency, easy to fall into local optimal solutions, lack of dynamic adjustment capabilities and poor adaptability to complex community structures in traditional community optimization methods, this scheme searches for high-quality community parameters through search initialization, definition of community quality evaluation function, design of dynamic guidance factors, dynamic evolutionary search, design of reset strategy and design of search rules, avoiding blind search, improving search efficiency, enhancing global search capabilities, enhancing the flexibility and adaptability of the search process, and ensuring the stability and reliability of the optimization results.
[0132] Example 6, see Figure 1 This embodiment is based on the above embodiment. The genetic care plan generation module inputs the new pregnant woman data into the constructed community model, reconstructs the genetic care community, obtains the community data to which the new pregnant woman data belongs, calculates the correlation between all data samples in the community and the new pregnant woman data, sets the care plan data of the data sample with the highest correlation as the care plan for the new pregnant woman, and provides genetic care for the pregnant woman.
[0133] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0134] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.
[0135] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.
Claims
1. Prenatal diagnosis genetic care system based on big data, characterized by: It includes data collection module, data expansion module, genetic care community construction module, community optimization module, and genetic care program generation module; The data collection module collects the prenatal diagnosis data and nursing plan data of the pregnant woman; The data expansion module performs data expansion by data partitioning, designing a data expansion model architecture, designing an adversarial loss function, designing a cyclic loss function, designing a feature loss function, and model comprehensive training; Design the adversarial loss function, which is expressed as follows: ； in, represents a first generator that converts source domain data into extended domain data, represents the first discriminator function, represents the adversarial loss function, Indicates extended domain data. Indicates the extended domain data distribution, represents the first discriminator function, represents the discrimination result of the first discriminator on the extended domain data, represents the logarithmic function, represents the expected output of the logarithm of the first discriminator’s discriminant result on the extended domain data, Represents source domain data, represents the source domain data distribution, represents the conditional vector, represents the data generated by combining the source domain data and the conditional vector, Represents the data generated by the first discriminator The expected output of the judgment result is to take the opposite number, add one, and then take the logarithm; The genetic care community building module generates a genetic care community by defining the similarity of prenatal diagnosis features, defining a basic gravity function, creating an initial gravity graph, data connection, setting an initial community center, calculating association, data point allocation, designing a community adjustment factor, and updating a community center; The community optimization module searches for high-quality parameters for the community by searching for initialization, defining a community quality evaluation function, designing a dynamic guiding factor, dynamic evolutionary search, designing a reset strategy, and designing search rules; The genetic care plan generation module inputs the new pregnant woman data into the constructed community model, calculates the correlation between all data samples in the community to which the new pregnant woman data belongs and the new pregnant woman data, sets the care plan data of the data sample with the highest correlation as the care plan for the new pregnant woman, and provides genetic care for the pregnant woman.
2. The prenatal diagnosis genetic care system based on big data according to claim 1 is characterized by: The data expansion module specifically includes the following contents: Data partitioning: creating two data domains: one is the source domain containing all original data, and the other is the extended domain for storing all expanded data. Design the data expansion model architecture, use the long short-term memory network as the basic architecture, and build two generators and one discriminator; the first generator includes an input layer, an LSTM layer, a fully connected layer, and an output layer; the LSTM layer is activated by a hyperbolic tangent function, and the fully connected layer is activated by a linear rectifier function; the second generator has the same structure as the first generator but in the opposite direction; the fully connected layer of the first discriminator is activated by a sigmoid function, and the rest of the structure is the same as the first generator; Design adversarial loss functions; Design the cycle loss function, which is expressed as follows: ； in, represents a second generator that converts the extended domain data into the source domain data, represents the cycle loss function, Represents source domain data After being converted to the extended domain by the first generator, the first reconstruction error is converted back to the source domain by the second generator. Indicates extended domain data After being converted to the source domain by the second generator, the second reconstruction error is converted back to the extended domain by the first generator; Design feature loss function, expressed as follows: ； in, represents the feature loss function, Indicates extended domain data Input to the first generator When the error between the output data and the original data is Represents source domain data Input to the second generator When , the error between the output data and the original data; Model comprehensive training, uses three loss functions to conduct comprehensive training of the data expansion model, which is expressed as follows: ； in, The comprehensive loss value of the model, and Indicates importance weight; Data expansion: Use the trained generator to generate prenatal diagnosis data and care plan data for pregnant women to expand the original dataset.
3. The prenatal diagnosis genetic care system based on big data according to claim 1, characterized in that: The genetic care community building module specifically includes the following: The similarity of prenatal diagnosis features is defined as follows: ； Among them, i and j represent the index of the data sample, and represent the i-th and j-th data samples respectively, represents the similarity of prenatal diagnosis features between the i-th and j-th data samples, f represents the index of prenatal diagnosis features, represents the total number of prenatal diagnostic features, represents the similarity weight corresponding to the f-th feature, represents the fth eigenvalue of the i-th data, represents the fth eigenvalue of the jth data, Indicates square root; Define the basic gravitational function as follows: ； in, represents the basic gravitational strength between the i-th and j-th data samples, e represents a natural constant, and represents the gravitational transformation factor; Create an initial gravity map, create an empty map , where D is the set of data samples and Q is the set of edges; Data connection, calculate the gravitational force between each pair of data samples, in the figure Add an edge connecting the data pair and set the gravity value as the weight of the edge; Set the initial community center and select data points as the initial community center; Calculate the relevance and calculate the relevance of each data point to each community center, as shown below: ； in, and Respectively represent the data points and community centers for calculating the association degree, Represents data points With community center The correlation between represents the weight of the edge between the data point and the community center, Represents data points The fth eigenvalue of Indicates community center The fth eigenvalue of Represents data points With community center Similarity of prenatal diagnosis characteristics between Data point allocation, where the data point is allocated to the community to which the community center with the greatest correlation with the data point belongs; Design the community adjustment factor, set the Euclidean distance between the two farthest data points as the community width, calculate the value obtained by dividing the total number of data points in the community by the community width, set this value as the density value of the community, and perform dynamic adjustment of the community, as shown below: ； in, represents the community adjustment factor, and represents the scale adjustment factor, Represents the total number of data points in the community, Indicates the number of data points whose distance from the community center is less than 0.3 times the community width. Indicates the current density, and denote the minimum and maximum community density, respectively; Community center update: calculate the mean of the positions of all data points in a community to obtain the average position of the community, fine-tune the average position, select the data point closest to the fine-tuned position, and set it as the latest community center, as shown below: ； in, Indicates the new community center, It means taking the nearest data point. represents the average location of the community, represents the community update weight, represents the symbolic function, represents the fine-tuning threshold, Indicates the old community center, Indicates taking the absolute value; Generate communities, iteratively assign data points and update community centers, and assign data points to communities until the communities converge and form a stable community structure.
4. The prenatal diagnosis genetic care system based on big data according to claim 1, characterized in that: The community optimization module specifically includes the following contents: Initialize the search, determine the search parameters, including the number of communities and the location of the initial community center, create a parameter search space, and generate an initial parameter search point cluster; The community quality evaluation function is defined as follows: ； in, Indicates the quality of the community, and They represent the community similarity weight and discreteness weight respectively, m represents the index of the community, and n represents the index of the data point in the community. and Represent the total number of communities and the total number of data points in the community, respectively. Indicates the Euclidean distance. represents the nth data point in the mth community, represents the mth community center, represents the m+1th community center; Design a dynamic guidance factor, expressed as follows: ； in, represents the dynamic guidance factor, t represents the current parameter search times, T represents the maximum parameter search times, , and represents the bootstrap constant, represents the Euclidean distance between the current search point and the search point with the highest community quality, and e represents a natural constant; Dynamic evolutionary search uses dynamic guiding factors to design parameter search strategies, which are expressed as follows: ； in, Indicates the position of the parameter point obtained by dynamic evolution search during the t+1th parameter search, Indicates the position of the parameter point obtained by dynamic evolution search during the tth parameter search, Represents a random number between 0 and 1. represents the average position of the search point cluster generated by the search initialization, is the position of the search point with the highest community quality among the searched parameter points, and check(·) means taking the nearest integer; Design a reset strategy and adjust the search direction through the reset strategy, as shown below: ； in, Indicates the position of the parameter point obtained by executing the reset strategy during the t+1th parameter search. represents the position of the parameter point obtained by executing the reset strategy during the t-th optimization search. Represents a random number ranging from 0 to 1; Design search rules, set the maximum number of parameter searches, set the community quality standard value, calculate the community quality corresponding to the initial parameter search point cluster, perform dynamic evolutionary search on the parameter search point cluster, calculate the community quality corresponding to the searched parameters, and if the community quality corresponding to more than 70% of the parameter points has been improved, perform the next dynamic evolutionary search, otherwise execute the reset strategy first, and then perform the next dynamic evolutionary search; if a parameter point with a community quality greater than the community quality standard value appears during the search process, output the parameters of the search point with the highest community quality among the parameter points; if the maximum number of parameter searches is reached, perform parameter search again; otherwise, continue searching.
5. The prenatal diagnosis genetic care system based on big data according to claim 1, characterized in that: The data collection module collects the prenatal diagnosis data and nursing plan data of the pregnant woman, wherein the prenatal diagnosis data includes serological test data, genetic test data, ultrasound examination data and family genetic disease history data; the nursing plan data includes life care measures, drug management measures, psychological care measures and education measures; The genetic care plan generation module inputs the new pregnant woman data into the constructed community model, reconstructs the genetic care community, obtains the community data to which the new pregnant woman data belongs, calculates the correlation between all data samples in the community and the new pregnant woman data, sets the care plan data of the data sample with the highest correlation as the care plan for the new pregnant woman, and provides genetic care for the pregnant woman.