Biomedical data statistics system and its analysis method

Abstract

The present invention relates to the field of bioinformatics technology, specifically to a biomedical data statistics system and an analysis method thereof, wherein the system comprises a data capture module, a structure analysis module, a manifold construction module, a global optimization module, a genetic network inference module, and a bioinformatics interpretation module. In the present invention, by finely capturing and characterizing biomedical data, genetic sequence data can be processed more effectively, the recognition accuracy of genetic features can be improved, and the local fit of data points and the detection of geometric structures can be used to clearly show the subtle differences between samples. The technology of simplifying multidimensional data into two-dimensional or three-dimensional representations helps to more accurately depict gene networks, thereby promoting an in-depth understanding of disease mechanisms and the discovery of potential rehabilitation methods, significantly improving the data processing efficiency and accuracy of biomedical research, and playing a positive role in promoting the development of precision medicine.

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics, and in particular to a biomedical data statistics system and an analysis method thereof. Background Art

[0002] Bioinformatics is an interdisciplinary subject that mainly uses computational technology to analyze biological data, especially large-scale genetic information. It covers a wide range of technologies from genome sequencing and alignment to protein structure analysis and function prediction. With the improvement of computing power and the advancement of biotechnology, bioinformatics has become an indispensable part of modern biological science and precision medicine. It analyzes the complexity of biological systems by establishing and applying various algorithms, computational models and statistical techniques in order to discover new biological knowledge or potential treatments.

[0003] Among them, biomedical data statistics systems refer to technical systems specifically used to collect, process and analyze data generated in biomedical research. The main purpose is to improve the efficiency and accuracy of data processing, support decision-making in diseases, treatment effect evaluation, and new drug development. Through advanced data statistics systems, researchers and medical professionals can better understand the mechanism of diseases and improve patients' treatment effects and quality of life.

[0004] Existing bioinformatics technologies have made remarkable achievements, but they still have limitations when processing large-scale genetic information, especially in terms of the multidimensionality and complexity of data. Traditional technologies are difficult to effectively reduce the information loss when converting high-dimensional data into low-dimensional representations, resulting in inaccurate data interpretation. Existing technologies also show certain deficiencies in identifying complex dependencies between genes, limiting the ability to deeply understand the mechanisms of diseases. For example, the lack of effective tools to distinguish between direct or indirect interactions between genes leads to insufficient identification of key genes and biomarkers when studying genetic diseases, affecting the formulation of treatment strategies and the development of new drugs. Insufficient treatments cause inaccuracies in dealing with actual biomedical problems, affecting the treatment effect and quality of life of patients. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a biomedical data statistics system and an analysis method thereof.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, the biomedical data statistics system comprises:

[0007] The data capture module extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, characterizes the data points according to the position variation frequency, and generates a data feature matrix;

[0008] The structural analysis module extracts data points from the data feature matrix, constructs an adjacency matrix of the data points, captures the local fit between samples, and obtains the local geometric structure;

[0009] The manifold construction module converts the multi-dimensional data into two-dimensional and three-dimensional spatial representations based on the local geometric structure, verifies the invariance of the distance between the original data points, and constructs a finite-dimensional manifold graph;

[0010] The global optimization module adjusts the position of the data points in the finite dimensional space according to the finite dimensional manifold diagram, avoids the information loss in the conversion from multi-dimensional to finite dimensional, and generates an optimized manifold structure;

[0011] The genetic network inference module uses the optimized manifold structure to analyze the mutual information between genes, and identifies the gene dependencies of direct and indirect relationships by quantifying the information mobility between genes, thereby obtaining a gene dependency graph;

[0012] The bioinformatics interpretation module analyzes the gene dependency graph to identify key genes and expression patterns in differentiated diseases, and obtains biomedical comparison results by comparing known disease markers.

[0013] As a further scheme of the present invention, the data feature matrix includes the position, type and variation frequency of nucleotides, the local geometric structure includes connection strength, connection type and connection distance, the finite dimensional manifold diagram includes two-dimensional and three-dimensional data space representations, a manifold diagram that keeps the distance between original data points unchanged, the spatial coordinates of points and the distance relationship between points, the optimized manifold structure includes the optimized position of data points in the finite dimensional space, the optimized degree of information loss, the optimized coordinate values ​​and the optimized distance error, the gene dependency graph includes the mutual information analysis results between genes, quantified information flow data, the mutual information scores of gene pairs and the interaction types and information flow directions, and the biomedical comparison results include a list of identified key genes, gene expression patterns in differentiated diseases and matching known disease markers.

[0014] As a further solution of the present invention, the data capture module includes:

[0015] The nucleotide location submodule extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, and encodes the sequence by identifying and recording the position information to obtain the nucleotide position index;

[0016] The variation frequency analysis submodule uses the nucleotide position index to analyze and compare the occurrence frequency of nucleotides, identify and record the frequency variation pattern, and obtain the position variation spectrum by comparing and analyzing the variation characteristics of the statistical position;

[0017] The matrix construction submodule encodes and sorts the features in the genetic sequence through the position variation spectrum, constructs the feature description of each data point using the variation data, and establishes a data feature matrix.

[0018] As a further solution of the present invention, the structural analysis module includes:

[0019] The data point extraction submodule selects key data points from the complete feature matrix, verifies the representativeness and relevance of each data point, and performs initial classification and labeling on the data points to generate a selected data point set;

[0020] The connection analysis submodule uses the screened data point set to calculate the relationship strength between the data points, and obtains the data point adjacency matrix by quantitatively analyzing the connection between each point and its neighboring points;

[0021] The fit analysis submodule evaluates and calculates the local fit between samples through the data point adjacency matrix, analyzes the consistency and variability of the local geometric structure, captures and describes the local structural characteristics between samples, and obtains the local geometric structure.

[0022] As a further solution of the present invention, the manifold building module includes:

[0023] The multidimensional data conversion submodule maps the multidimensional data into two-dimensional and three-dimensional space based on the local geometric structure, analyzes the spatial position of each data point, verifies the integrity of the data during the conversion process, and obtains a three-dimensional spatial representation;

[0024] The data distance verification submodule uses the three-dimensional space representation to analyze the Euclidean distance between the original data points, and obtains the distance verification result by comparing and analyzing the consistency of the distance before and after the data conversion;

[0025] The manifold graph construction submodule integrates data points according to the distance verification result, optimizes the topological structure and geometric property expression of the manifold graph, and constructs a finite-dimensional manifold graph.

[0026] As a further solution of the present invention, the global optimization module includes:

[0027] The data point adjustment submodule uses a manifold learning algorithm based on the finite dimensional manifold graph to adjust the position of the data point in the finite dimensional space to obtain the adjusted data point position;

[0028] The information loss avoidance submodule uses the adjusted data point positions to analyze and calculate the information loss in the process of converting from multi-dimensional to finite-dimensional, uses error correction and compensation techniques to optimize data representation, verifies the integrity of key information, and obtains information loss optimization results;

[0029] The consistency analysis submodule verifies the global consistency of the data and the matching degree of local features according to the information loss optimization result, analyzes the topological relationship of the data, and generates an optimized manifold structure.

[0030] As a further solution of the present invention, the formula of the manifold learning algorithm is as follows:

[0031] ;

[0032] in, represents the new position of data point i in the finite-dimensional space, N(i) is the number of data points, represents the position of data point i in the original multidimensional space, Represents the position of the neighbor j of data point i in the original multidimensional space, Represents the weight between data point i and its nearest neighbor j, absolute value Represents the distance between data point i and its nearest neighbor j in the original multidimensional space.

[0033] As a further solution of the present invention, the genetic network inference module includes:

[0034] The mutual information analysis submodule performs mutual information analysis between genes based on the optimized manifold structure, calculates the mutual information value of gene pairs, analyzes the covariance of gene expression, captures the potential communication mode between genes, and obtains the mutual information matrix of gene pairs;

[0035] The mobility quantification submodule uses the gene pair mutual information matrix to quantify the information mobility between genes, evaluate the propagation path and intensity of information in the gene network, and obtain information mobility analysis results;

[0036] The dependency identification submodule identifies the direct and indirect dependencies between genes according to the information mobility analysis results, distinguishes and annotates key nodes and connections in the network, and obtains a gene dependency graph.

[0037] As a further solution of the present invention, the bioinformation interpretation module includes:

[0038] The gene analysis submodule analyzes key genes in biological regulation and disease development according to the gene dependency graph, analyzes characteristics and functions, evaluates expression levels under differentiated conditions, and obtains key gene sets;

[0039] The expression pattern analysis submodule uses the key gene set to analyze the expression pattern of genes in differential diseases, and generates differential expression analysis results by comparing the expression differences between healthy and disease states;

[0040] The disease marker comparison submodule compares the known disease marker data according to the differential expression analysis results, captures the gene expression features of the same type and same as the known markers, verifies the correspondence between the identified patterns and the known pathological data, and obtains the biomedical comparison results.

[0041] A biomedical data statistical analysis method, which is performed based on the above-mentioned biomedical data statistical system, comprises the following steps:

[0042] S1: Extract genetic sequence data from biomedical data, mark the position of each nucleotide in the sequence, record the variation frequency of each position, and generate a nucleotide position marker map;

[0043] S2: Analyze the connection relationship between data points through the nucleotide position marker map, capture the local fit between samples, and generate an adjacency structure matrix;

[0044] S3: Based on the adjacency structure matrix, the multidimensional data is converted into a representation of two-dimensional and three-dimensional space, the invariance of the distance between the original data points is verified, and a finite-dimensional manifold graph is generated;

[0045] S4: using the finite-dimensional manifold graph, adjusting the position of the data points in the finite-dimensional space, avoiding information loss in the conversion from multi-dimensional to finite-dimensional, and generating an optimized manifold structure by measuring and adjusting the distance between the data points;

[0046] S5: Through the optimized manifold structure, mutual information analysis between genes is performed, and by quantifying the information flow between genes, the expression patterns of key genes and in differentiated diseases are analyzed, compared with known disease markers, and biomedical analysis results are generated.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are:

[0048] In the present invention, by finely capturing and characterizing biomedical data, genetic sequence data can be processed more effectively and the recognition accuracy of genetic characteristics can be improved. The local fit of data points and the detection of geometric structures are used to clearly show the subtle differences between samples, which are not easy to capture in traditional methods. The technology of simplifying multidimensional data into two-dimensional or three-dimensional representation effectively maintains the integrity of complex relationships between data and avoids the common information loss in the dimensionality reduction process. Through in-depth analysis of the mutual information between genes, the direct and indirect dependencies between genes are identified, which helps to more accurately depict the gene network, thereby promoting a deeper understanding of the disease mechanism and the discovery of potential rehabilitation methods, significantly improving the data processing efficiency and accuracy of biomedical research, and playing a positive role in promoting the development of precision medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a system flow chart of the present invention;

[0050] Figure 2 It is a schematic diagram of the system framework of the present invention;

[0051] Figure 3 It is a flow chart of the data capture module of the present invention;

[0052] Figure 4 It is a flow chart of the structural analysis module of the present invention;

[0053] Figure 5 It is a flow chart of the manifold building module of the present invention;

[0054] Figure 6 It is a flow chart of the global optimization module of the present invention;

[0055] Figure 7 It is a flow chart of the genetic network inference module of the present invention;

[0056] Figure 8 is a flow chart of the biological information interpretation module of the present invention;

[0057] Fig. 9 Schematic diagram of the analysis method steps of the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments thereof. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0059] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the 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 orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0060] See also Figure 1 to Figure 2 The present invention provides a technical solution, a biomedical data statistics system comprising:

[0061] The data capture module extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, characterizes the data points according to the position variation frequency, and generates a data feature matrix;

[0062] The structural analysis module extracts data points from the data feature matrix, constructs the adjacency matrix of the data points, captures the local fit between samples, and obtains the local geometric structure;

[0063] The manifold construction module converts multidimensional data into two-dimensional and three-dimensional spatial representations based on local geometric structures, verifies the invariance of the distance between original data points, and constructs finite-dimensional manifold graphs;

[0064] The global optimization module adjusts the position of data points in the finite-dimensional space according to the finite-dimensional manifold diagram, avoids information loss in the conversion from multi-dimensional to finite-dimensional, and generates an optimized manifold structure;

[0065] The genetic network inference module uses the optimized manifold structure to analyze the mutual information between genes, and by quantifying the information flow between genes, it identifies the gene dependencies of direct and indirect relationships and obtains the gene dependency graph;

[0066] The bioinformatics interpretation module analyzes the gene dependency graph to identify key genes and their expression patterns in differentiated diseases, and obtains biomedical comparison results by comparing known disease markers.

[0067] The data feature matrix includes the position, type and variation frequency of nucleotides. The local geometric structure includes connection strength, connection type and connection distance. The finite-dimensional manifold diagram includes two-dimensional and three-dimensional data space representations, manifold diagrams that keep the distance between original data points unchanged, the spatial coordinates of points and the distance relationship between points. The optimized manifold structure includes the optimized position of data points in the finite-dimensional space, the optimized degree of information loss, the optimized coordinate values ​​and the optimized distance error. The gene dependency diagram includes the results of mutual information analysis between genes, quantitative information mobility data, mutual information scores of gene pairs and interaction types and information flow directions. The biomedical comparison results include a list of identified key genes, gene expression patterns in differentiated diseases and matching known disease markers.

[0068] See also Figure 2 , 3 , the data capture module includes:

[0069] The nucleotide location submodule extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, and encodes the sequence by identifying and recording the position information to obtain the execution process of the nucleotide position index as follows;

[0070] The extraction of genetic sequence data is the basis. Nucleotide coding technology is used to accurately mark the position of each nucleotide in the sequence. The position information is recorded through high-precision coding technology to ensure the accuracy and integrity of the data. It not only marks the nucleotide position, but also ensures the traceability and reusability of the data through the coding process. The accurate recording of position information provides a solid foundation for the next step of data analysis. By accurately encoding the genetic sequence, high-quality basic data is provided for subsequent data analysis and research, ensuring the accuracy and efficiency of data processing, and providing scientific researchers with a reliable data support, so that research work can be carried out based on accurate basic data, increasing the depth and breadth of the research, and obtaining a nucleotide position index.

[0071] The variation frequency analysis submodule uses nucleotide position indexes to analyze and compare the occurrence frequencies of nucleotides, identify and record the frequency variation patterns, and obtain the position variation spectrum by comparing and analyzing the variation characteristics of the statistical positions. The execution process is as follows;

[0072] By comparing and analyzing the variation characteristics of the statistical positions through the nucleotide position index, according to the formula , calculate the position variation spectrum. Where f(x) represents the position variation spectrum, represents the nucleotide variation status at position a, An indicator function indicating whether there is a mutation at position a, Represents the total number of positions in the nucleotide sequence.

[0073] Detailed explanation of the formula and the process of formula calculation: Consider a specific genetic sequence with 100 positions, where positions 2, 5, and 10 are mutated. Set the indicator function when The position has variation, otherwise it is 0. Therefore, , , , the remaining positions Substituting the value into the above formula, we get: The calculation process shows that the value of the position variation spectrum is 0.03, indicating that about 3% of the 100 nucleotide positions have variations. The results are helpful for further analysis of the variation pattern and biological significance of the sequence.

[0074] The matrix construction submodule encodes and sorts the features in the genetic sequence through the position variation spectrum, and uses the variation data to construct the feature description of each data point. The execution process of establishing the data feature matrix is ​​as follows;

[0075] The use of position variation spectrum is the key. The features in the genetic sequence are encoded and sorted through the position variation spectrum. The construction of the feature description of each data point involved in the construction of variation data is a refined operation. Statistical methods are needed to evaluate the influence of each variation position. By sorting and encoding position variation data, not only important genetic variations can be identified, but also valuable insights can be provided for genetic research. Accurate execution is achieved by calculating the statistical values ​​corresponding to each variation position. Establishing a data feature matrix is ​​a complex process involving multi-level data processing. In bioinformatics analysis and medical research, it provides support for disease prediction and provides important tools for subsequent data processing and analysis to establish a data feature matrix.

[0076] See also Figure 2 , 4 , the structural analysis module includes:

[0077] The data point extraction submodule selects key data points from the complete feature matrix, verifies the representativeness and relevance of each data point, and initializes the classification and labeling of the data points. The execution process of generating the selected data point set is as follows;

[0078] Screening key data points from the complete feature matrix involves complex data analysis techniques and algorithms. The key to screening is to determine the representativeness and relevance of each data point. Through in-depth analysis of the complete feature matrix, select data points that can represent the overall data characteristics or play an important role in specific analysis. The process not only needs to use statistical analysis techniques to evaluate the importance of data points, but also needs to verify the actual application value of data points through machine learning models. The screened data points need to be preliminarily classified and labeled in preparation for subsequent data processing and analysis. They can effectively represent the important characteristics of the overall data set, provide an accurate input source for subsequent analysis, and generate a screened data point set.

[0079] The connection analysis submodule uses the selected data point set to calculate the relationship strength between data points. By quantitatively analyzing the connection between each point and its neighboring points, the execution process of the data point adjacency matrix is ​​obtained as follows;

[0080] The core task is to use the screened data point set to construct an adjacency matrix between data points. By calculating the strength of the relationship between each data point, a comprehensive data connection network is formed. The adjacency matrix reflects the interaction and connection between data points. The quantitative analysis of this relationship is completed through a variety of mathematical models and algorithms, including distance measurement and similarity calculation, which can clarify the connection between each data point and its neighboring points. This matrix plays an important role in subsequent data analysis and model building, providing a methodological foundation so that the structure and relationship of data points can be clearly expressed mathematically and the data point adjacency matrix can be obtained.

[0081] The fit analysis submodule evaluates and calculates the local fit between samples through the data point adjacency matrix, analyzes the consistency and variability of the local geometric structure, captures and describes the local structural characteristics between samples, and obtains the execution process of the local geometric structure as follows;

[0082] The local geometric structure consistency and variability between samples are analyzed through the data point adjacency matrix, according to the formula , calculate the local fit. Where C(x) represents the local fit, and is the neighboring point of point x in the adjacency matrix, b and c represent the grouping labels of different data points, is the neighborhood set of point x.

[0083] Detailed explanation of the formula and the process of formula calculation: Assume that a point x in an adjacency matrix has three neighboring points, and the eigenvalues ​​of the points are 1, 3, and 5 respectively. Calculate the local fit according to the formula:

[0084] ;

[0085] Calculations show that the local fit value is 24. The value reflects the sum of the squares of the differences in the eigenvalues ​​between point x and its neighboring points. Higher values ​​indicate greater local structural differences, which can be used to evaluate the consistency and variability between samples. This is of great significance for understanding and describing the local structural characteristics between samples.

[0086] See also Figure 2 , 5 , the manifold building blocks include:

[0087] The multidimensional data conversion submodule maps the multidimensional data into two-dimensional and three-dimensional space based on the local geometric structure, analyzes the spatial position of each data point, verifies the integrity of the data during the conversion process, and obtains the execution process of the three-dimensional spatial representation as follows;

[0088] Based on the local geometric structure, multidimensional data is mapped to two-dimensional and three-dimensional space. The process involves advanced data conversion technology and spatial analysis methods. Effectively reducing multidimensional data to a lower-dimensional space is a key step in achieving data visualization and further analysis. Through this conversion, the spatial position of each data point can be clearly expressed, which not only helps to reveal the inherent structure of the data, but also can intuitively display the relative position relationship between data points. Verification of data integrity is an important aspect of the process. It is necessary to check whether the data is lost or distorted during the conversion process to ensure that the converted data still retains the basic characteristics and structural information of the original data, so as to provide an accurate spatial model for subsequent data analysis and obtain a three-dimensional spatial representation.

[0089] The data distance verification submodule uses three-dimensional space representation to analyze the Euclidean distance between the original data points. By comparing and analyzing the consistency of the distance before and after the data transformation, the execution process of obtaining the distance verification result is as follows;

[0090] By comparing and analyzing the data before and after the transformation, the consistency of the distance is verified. According to the formula , calculate the data distance verification result. In the formula, d(x,y) represents the Euclidean distance between point x and point y, and are the coordinates of point x and point y in three-dimensional space respectively.

[0091] Detailed explanation of the formula and the process of formula calculation: Assume that the coordinates of point x are (1, 2, 3) and the coordinates of point y are (4, 6, 8). Substitute them into the formula to calculate the Euclidean distance:

[0092] .

[0093] The calculation process shows that the Euclidean distance between point x and point y is approximately 7.07 units of length. This precise distance calculation helps verify the accuracy of data conversion and ensure the consistency and reliability of three-dimensional spatial representation.

[0094] The manifold graph construction submodule integrates data points based on the distance verification results, optimizes the topological structure and geometric property expression of the manifold graph, and constructs a finite-dimensional manifold graph. The execution process is as follows;

[0095] Based on the distance verification results, integrating data points and optimizing the topological structure of the manifold graph are the core tasks. By adjusting the connections and relationships between data points, the expression of the manifold graph is optimized, which involves complex topological optimization technology and geometric property analysis, which can better express the intrinsic structure and characteristics of the data. The optimized manifold graph not only has better geometric property expression, but also reflects the actual physical or abstract relationship of the data. The construction of the manifold graph not only depends on mathematics and computational methods, but also requires comprehensive data The actual application scenarios and needs are needed to ensure the practicality of the manifold graph and the accuracy of information expression, providing powerful tools and methods for data analysis and further research, and constructing finite-dimensional manifold graphs.

[0096] See also Figure 2 , 6 , the global optimization module includes:

[0097] The data point adjustment submodule is based on a finite-dimensional manifold graph and uses a manifold learning algorithm to adjust the position of the data point in the finite-dimensional space. The execution process of obtaining the adjusted data point position is as follows;

[0098] The formula for the manifold learning algorithm is as follows:

[0099] ;

[0100] in, Representative data points The new position in the finite-dimensional space, N(i) is the number of data points, represents the position of data point i in the original multidimensional space, Representative data points Neighbors The position in the original multidimensional space, Represents the weight between data point i and its nearest neighbor j, absolute value Represents the distance between data point i and its nearest neighbor j in the original multidimensional space.

[0101] Detailed explanation of the formula and the process of formula calculation and derivation:

[0102] In the local linear embedding algorithm, the new position of each data point i is In the finite dimensional space, by considering the nearest neighbors in the original high dimensional space to calculate the position.

[0103] The process includes the following steps:

[0104] 1. Calculate the relationship between data point i and each neighbor The Euclidean distance between . It is obtained by directly measuring the coordinates of the data points in the high-dimensional space. For example, if the data point i is located at (2, 3) and its neighbors Located at (5, 7), the distance is

[0105] .

[0106] Determine the weight , the weight is obtained by minimizing the reconstruction error between each point and its neighbors. For example, if the similarity between neighbors is high, the weight is large, otherwise it is small. To ensure the rationality of the weight, it can be obtained through expert data or historical data analysis, such as by comparing the effects of different strategies in experiments to determine the optimal weight. If the similarity between i and j is 0.8, then Set to 0.8.

[0107] 2. Calculate new coordinates The numerator of is the sum of the weighted distances of all neighbors to i. According to the above values, if There are three nearest neighbors with weights of 0.8, 0.5, and 0.3, and distances of 5, 3, and 4, respectively. The numerator is 0.8×5+0.5×3+0.3×4=4+1.5+1.2=6.7.

[0108] 3. Calculate the denominator, which is the square root of the sum of the squares of the weights. Taking the above weights as an example, the denominator is

[0109] ;

[0110] 4. Through calculation, we get The value of , the value represents the new position of the data point i in the low-dimensional space after adjustment.

[0111] The results show that through reasonable weight allocation and accurate calculation of the distance between adjacent data points, the intrinsic structure of high-dimensional data can be effectively mapped in low-dimensional space, the distribution between data points can be optimized, and the parsing ability and visualization effect of the model can be improved.

[0112] The information loss avoidance submodule uses the adjusted data point positions to analyze and calculate the information loss during the conversion from multi-dimensional to finite-dimensional, uses error correction and compensation techniques to optimize data representation, verifies the integrity of key information, and obtains the information loss optimization results. The execution process is as follows;

[0113] Analyze and calculate the information loss in the process of transformation from multi-dimensional to finite-dimensional, according to the formula

[0114] , and get the information loss optimization result. In the formula, represents the information loss ratio, is the variance contribution of the dth principal component, is the number of principal components retained in the finite dimensional space, and m is the total dimension of the original data.

[0115] Detailed explanation of the formula and the process of formula calculation: Set the total dimension m of the original data to 100, and the retained dimension is 20, and the variance contributions of each dimension are , for all d, substitute into the formula to calculate:

[0116] ;

[0117] Calculations show that 80% of the information is lost during the conversion process. The results help to measure the degree of information loss during dimensionality reduction and optimize data representation through further error correction and compensation techniques.

[0118] The consistency analysis submodule verifies the global consistency of the data and the matching degree of local features based on the information loss optimization results, analyzes the topological relationship of the data, and generates the optimized manifold structure. The execution process is as follows;

[0119] The global consistency of the data and the matching degree of local features are verified based on the information loss optimization results. The process involves complex data analysis techniques, especially in analyzing the topological relationship of the data. Detailed data structure analysis ensures that the essential characteristics of the data can be accurately reflected, which not only retains the key information of the data, but also improves the interpretability and visualization quality of the data. Researchers can obtain an optimized manifold that reflects both global characteristics and local data characteristics, providing a powerful tool for in-depth analysis and further research, and generating an optimized manifold structure.

[0120] See also Figure 2 , 7 , the genetic network inference module includes:

[0121] The mutual information analysis submodule performs mutual information analysis between genes based on the optimized manifold structure, calculates the mutual information value of gene pairs, analyzes the covariance of gene expression, captures the potential communication pattern between genes, and obtains the execution process of the mutual information matrix of gene pairs as follows;

[0122] The optimized manifold structure is used to perform mutual information analysis between genes. The process includes calculating the mutual information values ​​of gene pairs to analyze the covariance of gene expression, which can capture the potential communication patterns between genes. The calculation of mutual information values ​​reveals how genes influence each other during the expression process and reflects the synchronization and correlation of gene expression. This information is the key to understanding gene functions and interactions, and provides an important data basis for further biological research and genetic network analysis, and obtains the mutual information matrix of gene pairs.

[0123] The mobility quantification submodule uses the gene pair mutual information matrix to quantify the information mobility between genes and evaluate the propagation path and intensity of information in the gene network. The execution process of obtaining the information mobility analysis results is as follows;

[0124] By using the gene pair mutual information matrix, the information flow between genes is quantified, including evaluating the propagation path and intensity of information in the gene network, helping researchers understand how genes transmit information in the network and revealing the dynamics and complexity of information propagation. The quantification of information flow provides important insights into the structure and function of gene networks. The results are crucial for discovering new biological mechanisms and potential therapeutic targets, and can guide future genetic research and clinical applications to obtain information flow analysis results.

[0125] The dependency identification submodule identifies the direct and indirect dependencies between genes based on the information mobility analysis results, distinguishes and annotates the key nodes and connections in the network, and obtains the gene dependency graph. The execution process is as follows;

[0126] According to the information mobility analysis results, the direct and indirect dependencies between genes are identified, and the formula , and obtain the gene dependency graph. In the formula, represents the dependency coefficient between gene g and gene h, is the mutual information value between gene g and gene h, and are the entropy values ​​of gene g and gene h respectively.

[0127] Detailed explanation of the formula and the process of formula calculation: Set the mutual information value of gene g and gene h to 0.8, and their respective entropy values ​​to 2.0 and 2.5. Substitute into the formula for calculation:

[0128] ;

[0129] Calculations showed that there is a moderately strong dependency between gene g and gene h, and the results help scientists better understand how genes interact in networks, as well as their roles and importance in biological networks.

[0130] See also Figure 2 , 8 , the bioinformatics interpretation module includes:

[0131] The gene analysis submodule analyzes key genes in biological regulation and disease development according to the gene dependency graph, analyzes characteristics and functions, and evaluates expression levels under differentiated conditions to obtain the execution process of the key gene set as follows;

[0132] According to the gene dependency graph, key genes in biological regulation and disease development are analyzed. The process involves a detailed analysis of the characteristics and functions of genes and the evaluation of gene expression levels under different conditions. Genes that play a key role in biological regulation can be identified, and genes that are closely related to disease development can also be identified. Key gene sets provide important targets for further biomedical research and potential treatment strategy development, enabling researchers to conduct in-depth gene function research on specific biological processes or disease states and obtain key gene sets.

[0133] The expression pattern analysis submodule uses key gene sets to analyze the expression patterns of genes in differential diseases. By comparing the expression differences between healthy and disease states, the execution process of generating differential expression analysis results is as follows;

[0134] By comparing the expression differences between healthy and disease states, according to the formula , generate differential expression analysis results.

[0135] In the formula, represents the average value of expression difference, Is the gene in disease or health The expression level of is the total number of genes in the key gene set.

[0136] Detailed explanation of the formula and the process of formula calculation and derivation:

[0137] There are 5 key genes, whose expression levels in disease state are 10, 15, 20, 25, 30, and in healthy state are 5, 10, 15, 20, 25. Substitute into the formula for calculation: ;

[0138] The calculation results show that the average expression level of genes in the disease state is 5 units higher than that in the healthy state. Differential expression analysis helps to reveal how genes regulate their functions under different health conditions and provide key data for diseases.

[0139] The disease marker comparison submodule compares the known disease marker data based on the differential expression analysis results, captures the gene expression features of the same type and the same as the known markers, verifies the correspondence of the identified patterns with the known pathological data, and obtains the execution process of the biomedical comparison results as follows;

[0140] Based on the results of differential expression analysis, comparison with known disease marker data is performed to capture gene expression features that are identical or similar to known markers. Through this comparison, the correspondence between the identified expression patterns and known pathological data can be verified, which is crucial for confirming the accuracy and relevance of research results, evaluating the disease relevance of newly discovered gene expression features, further ensuring the biomedical value and application prospects of gene markers, promoting the construction of a bridge between genetic research and clinical applications, and obtaining biomedical comparison results.

[0141] See also Fig. 9 , a biomedical data statistical analysis method, the biomedical data statistical analysis method is performed based on the above biomedical data statistical system, and includes the following steps:

[0142] S1: Extract genetic sequence data from biomedical data, mark the position of each nucleotide in the sequence, record the variation frequency of each position, and generate a nucleotide position marker map;

[0143] S2: Analyze the connection relationship between data points through the nucleotide position marker map, capture the local fit between samples, and generate an adjacency structure matrix;

[0144] S3: Based on the adjacency structure matrix, the multidimensional data is converted into two-dimensional and three-dimensional space representations, the invariance of the distance between the original data points is verified, and a finite-dimensional manifold graph is generated;

[0145] S4: Using the finite-dimensional manifold graph, adjust the position of data points in the finite-dimensional space to avoid information loss in the conversion from multidimensional to finite-dimensional space, and generate an optimized manifold structure by measuring and adjusting the distance between data points;

[0146] S5: By optimizing the manifold structure, information analysis is performed between genes. By quantifying the information flow between genes, key genes and expression patterns in differentiated diseases are analyzed, compared with known disease markers, and biomedical analysis results are generated.

[0147] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. Biomedical data statistics system, characterized in that: The system comprises: The data capture module extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, characterizes the data points according to the position variation frequency, and generates a data feature matrix; The structural analysis module extracts data points from the data feature matrix, constructs an adjacency matrix of the data points, captures the local fit between samples, and obtains the local geometric structure; The manifold construction module includes a multi-dimensional data conversion submodule, a data distance verification submodule and a manifold graph construction submodule; The multidimensional data conversion submodule maps the multidimensional data to two-dimensional and three-dimensional space based on the local geometric structure, analyzes the spatial position of each data point, verifies the integrity of the data during the conversion process, and obtains a three-dimensional spatial representation; The data distance verification submodule uses the three-dimensional space representation to analyze the Euclidean distance between the original data points, and obtains the distance verification result by comparing and analyzing the consistency of the distance before and after the data conversion; The manifold graph construction submodule integrates data points according to the distance verification result, optimizes the topological structure and geometric property expression of the manifold graph, and constructs a finite-dimensional manifold graph; The global optimization module adjusts the position of the data points in the finite dimensional space according to the finite dimensional manifold diagram, avoids the information loss in the conversion from multi-dimensional to finite dimensional, and generates an optimized manifold structure; The genetic network inference module uses the optimized manifold structure to analyze the mutual information between genes, and identifies the gene dependencies of direct and indirect relationships by quantifying the information mobility between genes, thereby obtaining a gene dependency graph; The genetic network inference module includes a mutual information analysis submodule, a mobility quantification submodule and a dependency identification submodule: The mutual information analysis submodule performs mutual information analysis between genes based on the optimized manifold structure, calculates mutual information values ​​of gene pairs, analyzes covariance of gene expression, captures potential communication patterns between genes, and obtains a mutual information matrix of gene pairs; The mobility quantification submodule uses the gene pair mutual information matrix to quantify the information mobility between genes, evaluate the propagation path and intensity of information in the gene network, and obtain information mobility analysis results; The dependency identification submodule identifies direct and indirect dependencies between genes based on the information mobility analysis results, distinguishes and annotates key nodes and connections in the network, and obtains a gene dependency graph; The bioinformatics interpretation module analyzes the gene dependency graph to identify key genes and expression patterns in differentiated diseases, and obtains biomedical comparison results by comparing known disease markers; The data feature matrix includes the position, type and variation frequency of nucleotides, the local geometric structure includes connection strength, connection type and connection distance, the finite-dimensional manifold diagram includes two-dimensional and three-dimensional data space representations, a manifold diagram that keeps the distance between original data points unchanged, the spatial coordinates of points and the distance relationship between points, the optimized manifold structure includes the optimized position of data points in the finite-dimensional space, the optimized degree of information loss, the optimized coordinate values ​​and the optimized distance error, the gene dependency diagram includes the mutual information analysis results between genes, quantified information flow data, the mutual information scores of gene pairs, the interaction types and the information flow direction, and the biomedical comparison results include a list of identified key genes, gene expression patterns in differentiated diseases and matched known disease markers.

2. The biomedical data statistics system according to claim 1, characterized in that: The data capture module comprises: The nucleotide location submodule extracts genetic sequence data from biomedical data, uses nucleotide encoding technology to mark the position of each nucleotide in the sequence, and encodes the sequence by identifying and recording the position information to obtain the nucleotide position index; The variation frequency analysis submodule uses the nucleotide position index to analyze and compare the occurrence frequency of nucleotides, identify and record the frequency variation pattern, and obtain the position variation spectrum by comparing and analyzing the variation characteristics of the statistical position; The matrix construction submodule encodes and sorts the features in the genetic sequence through the position variation spectrum, constructs the feature description of each data point using the variation data, and establishes a data feature matrix.

3. The biomedical data statistics system according to claim 1, characterized in that: The structural analysis module includes: The data point extraction submodule selects key data points from the data feature matrix, verifies the representativeness and relevance of each data point, and performs initial classification and labeling on the data points to generate a selected data point set; The connection analysis submodule uses the screened data point set to calculate the relationship strength between the data points, and obtains the data point adjacency matrix by quantitatively analyzing the connection between each point and its neighboring points; The fit analysis submodule evaluates and calculates the local fit between samples through the data point adjacency matrix, analyzes the consistency and variability of the local geometric structure, captures and describes the local structural characteristics between samples, and obtains the local geometric structure.

4. The biomedical data statistics system according to claim 1, characterized in that: The global optimization module includes: The data point adjustment submodule uses a manifold learning algorithm based on the finite dimensional manifold graph to adjust the position of the data point in the finite dimensional space to obtain the adjusted data point position; The information loss avoidance submodule uses the adjusted data point positions to analyze and calculate the information loss in the process of converting from multi-dimensional to finite-dimensional, uses error correction and compensation techniques to optimize data representation, verifies the integrity of key information, and obtains information loss optimization results; The consistency analysis submodule verifies the global consistency of the data and the matching degree of local features according to the information loss optimization result, analyzes the topological relationship of the data, and generates an optimized manifold structure.

5. The biomedical data statistics system according to claim 4, characterized in that: The formula of the manifold learning algorithm is as follows: ； in, represents the new position of data point i in the finite-dimensional space, is the number of data points, represents the position of data point i in the original multidimensional space, Represents the position of the neighbor j of data point i in the original multidimensional space, Represents the weight between data point i and its nearest neighbor j, absolute value Represents the distance between data point i and its nearest neighbor j in the original multidimensional space.

6. The biomedical data statistics system according to claim 1, characterized in that: The biological information interpretation module includes: The gene analysis submodule analyzes key genes in biological regulation and disease development according to the gene dependency graph, analyzes characteristics and functions, evaluates expression levels under differentiated conditions, and obtains key gene sets; The expression pattern analysis submodule uses the key gene set to analyze the expression pattern of genes in differential diseases, and generates differential expression analysis results by comparing the expression differences between healthy and disease states; The disease marker comparison submodule compares the known disease marker data according to the differential expression analysis results, captures the gene expression features of the same type and same as the known markers, verifies the correspondence between the identified patterns and the known pathological data, and obtains the biomedical comparison results.

7. A biomedical data statistical analysis method, characterized in that: The method is performed by the biomedical data statistics system according to any one of claims 1 to 6, comprising the following steps: Extract genetic sequence data from biomedical data, use nucleotide encoding technology to mark the position of each nucleotide in the sequence, characterize the data points according to the position variation frequency, and generate a data feature matrix; Extracting data points from the data feature matrix, constructing an adjacency matrix of the data points, capturing the local fit between samples, and obtaining a local geometric structure; Based on the local geometric structure, the multidimensional data is mapped to two-dimensional and three-dimensional space, the spatial position of each data point is analyzed, the integrity of the data in the transformation process is verified, and a three-dimensional space representation is obtained; the Euclidean distance between the original data points is analyzed using the three-dimensional space representation, and the consistency of the distance before and after the data transformation is verified by comparative analysis to obtain a distance verification result; according to the distance verification result, the data points are integrated, the topological structure and the expression of the geometric properties of the manifold graph are optimized, and a finite-dimensional manifold graph is constructed; According to the finite-dimensional manifold graph, the positions of the data points in the finite-dimensional space are adjusted to avoid information loss in the conversion from multi-dimensional to finite-dimensional, and an optimized manifold structure is generated; The optimized manifold structure is used to analyze the mutual information between genes, and by quantifying the information mobility between genes, the gene dependency relationships of direct and indirect relationships are identified to obtain a gene dependency graph; The bioinformatics interpretation module analyzes the gene dependency graph to identify key genes and expression patterns in differentiated diseases, and obtains biomedical comparison results by comparing known disease markers.

Images