Hepatoencephalopathy identification method, device and equipment based on metabolic connectivity omics and medium
By introducing connectomics indicators and individualized quantitative technology, the problem of insufficient accuracy of the identification method of hepatic encephalopathy is solved, and the construction of metabolic connectomics model and accurate classification of disease states is achieved at the individual level.
Patent Information
- Application Number
- CN202510389107.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-31
- Publication Date
- 2025-07-29
AI Technical Summary
The existing hepatic encephalopathy recognition methods rely on neuropsychological testing and imaging methods. The objectivity and accuracy are insufficient, making it difficult to achieve early accurate judgments. The revealing of the interaction between metabolites in complex metabolic networks is limited and quantitative analysis cannot be carried out at the individual level.
Connectionomics indicators are introduced, and through high-differential connection screening, key node screening and individualized quantitative technologies, the connection changes between metabolites are captured, and an individualized metabolic connectionomics model is constructed to improve the accuracy of disease state classification.
It significantly improves the accuracy of hepatic encephalopathy identification and personalized application capabilities, and achieves more accurate disease status classification and personalized medical strategies.
Smart Images
Figure CN120388729A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of metabolic connectomics, and in particular to a method, device, equipment and medium for identifying hepatic encephalopathy based on metabolic connectomics. Background Art
[0002] Hepatic encephalopathy is a neurological symptom caused by severe liver dysfunction and is a high-incidence complication of liver cirrhosis, seriously affecting the survival and quality of life of patients. The pathogenesis of hepatic encephalopathy is complex and involves disorders of multiple metabolic pathways. Existing examination methods rely on neuropsychological tests, clinical symptoms, biochemical indicators such as blood ammonia, and imaging means such as functional magnetic resonance imaging. However, these methods have problems such as insufficient objectivity and accuracy, and it is difficult to achieve early and accurate judgment of hepatic encephalopathy.
[0003] As an important branch of systems biology, metabolomics is committed to studying the changes of small molecule metabolites in organisms and the relationships between these changes and gene expression, protein function, and external environmental factors. In recent years, with the rapid development of high-throughput sequencing technology and mass spectrometry analysis technology, metabolomics has been increasingly widely used in disease diagnosis, drug screening, and personalized medicine. Hepatic encephalopathy, as a neurological manifestation of systemic metabolic disorders, can capture information helpful for disease identification and risk prediction through metabolomics. However, traditional metabolomics analysis techniques mainly focus on the up-regulation or down-regulation changes of single or a group of metabolites at the "level", with limited ability to reveal the interactions between metabolites in complex metabolic networks and insufficient understanding of the changes in network connection patterns under different disease states, and a large amount of valuable information has not been effectively extracted.
[0004] Connectomics is an omics technology originating from neuroscience, mainly exploring the connection relationships between nodes through network analysis. This idea also has a guiding role in mining metabolomics information. However, connectomics methods face challenges when applied to metabolomics data, especially in the quantification of metabolic connections at the individual level. Because the metabolomics data of a single individual cannot calculate the correlation coefficient matrix, which limits the application of connectomics technology at the individual level. In view of this, the present invention is specifically proposed. Summary of the Invention
[0005] The purpose of the present invention is to overcome the above technical deficiencies and provide a method, device, equipment and medium for identifying hepatic encephalopathy based on metabolic connectomics, so as to solve the technical problems in the related art that the ability to reveal the interactions between metabolites in complex metabolic networks is limited, the understanding of the changes in network connection patterns under different disease states is insufficient, and the metabolic connections at the individual level cannot be quantified, resulting in insufficient objectivity and accuracy of the hepatic encephalopathy identification method.
[0006] Different from traditional methods that only focus on the up - or down - regulation changes of metabolites, the present invention can capture the connection changes between metabolites by introducing "connectomics" indicators, and use individualized quantitative techniques to achieve the analysis of metabolic connectomics indicators and precise modeling at the individual level, significantly improving the accuracy of disease state classification.
[0007] To achieve the above technical objectives, the present invention adopts the following technical solutions: According to one aspect of the present invention, there is provided a method for identifying encephalopathy in hepatic encephalopathy metabolic connectomics, including: Step S1: Data acquisition and cleaning; the data acquisition and cleaning specifically include: acquiring a metabolite data set, and performing logarithmic transformation on the metabolic index data in the metabolite data set; Step S2: Screening of highly differential connections; The screening of highly differential connections includes: Step S21, calculating the Pearson correlation coefficient between each pair of all transformed metabolite indicators, respectively for the hepatic encephalopathy group and the control group; Step S22, comparing the correlation coefficients between the two groups. If the positive and negative signs are opposite, it is confirmed as a directional difference. If the signs are the same and the difference in the correlation coefficients is greater than or equal to a preset threshold, it is confirmed as a strong difference, otherwise it is a weak difference; Step S23, outputting the metabolite pairs with the directional difference and the strong difference, and confirming them as highly differential connections; Step S3: Screening of key nodes; the screening of key nodes includes: Step S31, disassembling the highly differential connections, and splitting each connection into two nodes; Step S32, mixing the disassembled nodes; Step S33, counting the frequencies of different metabolites as nodes, and extracting the top n nodes with the highest frequencies as key nodes; Step S4: Drawing of benchmark patterns; the drawing of benchmark patterns includes: Step S41, selecting the key nodes, and repeating Step S21 to calculate the correlation matrix for the hepatic encephalopathy group and the control group respectively; Step S42, drawing a heat map based on the correlation matrix to obtain two different connection patterns for hepatic encephalopathy and the control, respectively serving as the heat - related benchmark matrix and the cold - related benchmark matrix; Step S5: Individual connectomics quantification; the individual connectomics quantification includes: Step S51, respectively calculating the percentiles of the key nodes contained in the pattern benchmarks for the hepatic encephalopathy group and the control group to generate the hepatic encephalopathy position samples and the control position samples; Step S52: Combine the individual data with the hepatic encephalopathy location samples, and repeat Step S21 to obtain a heat-related mapping matrix; Step S53: Combine the individual data with the control location samples, and repeat Step S21 to obtain a cold-related mapping matrix; Step S54: Based on Step S52 and Step S53, record the correlation coefficients between pairwise key nodes as heat connection indicators and cold connection indicators respectively.
[0008] Optionally, the method further includes: Step S6: Pattern recognition based on metabolomics connectomics; the pattern recognition based on metabolomics connectomics includes: Step S61: Draw graphs based on the heat-related mapping matrix and the cold-related mapping matrix to visualize the pattern similarity with the heat-related reference matrix and the cold-related reference matrix.
[0009] Optionally, Step S6 further includes: Step S62: For cases that are not easily distinguishable, calculate the similarities between the heat-related mapping matrix and the cold-related mapping matrix and the heat-related reference matrix and the cold-related reference matrix respectively, and then complete pattern recognition, that is, disease state classification.
[0010] Optionally, to calculate the similarities between the heat-related mapping matrix and the cold-related mapping matrix and the heat-related reference matrix and the cold-related reference matrix, the Frobenius norm distance can be used. The smaller the distance, the more similar. The calculation formula is: where \(C\) is the mapping matrix to be determined, and \(A\) is the reference matrix.
[0011] Optionally, when the number of key nodes < 10, other methods such as the trace norm (the smaller the more similar) and the trace inner product (the larger the more similar) can also be used for the similarity evaluation method (other methods are also available).
[0012] Optionally, the method further includes: Step S7: Prediction based on metabolomics connectomics indicators; the prediction based on metabolomics connectomics indicators includes: Step S71: Screen the heat connection indicators and the cold connection indicators; Step S72: Based on the screened connection indicators, construct a prediction model.
[0013] Optionally, the screening methods for screening the heat connection indicators and the cold connection indicators include AUC, Lasso, elastic net, random forest variable importance, SHAP.
[0014] Optionally, the construction of the prediction model includes the random forest method.
[0015] The random forest method is to use the selected connection metrics as the variable set and whether it is hepatic encephalopathy as the classification result to combine and generate a data set.
[0016] The modeling method is to randomly select samples from the data set with replacement for training to construct decision trees. When splitting each time in the decision tree, a part of the variable set is randomly selected to avoid overfitting of the model. Multiple decision trees are integrated to form a random forest, and the final prediction result is determined by the highest frequency classification among all decision trees.
[0017] Optionally, during the construction of the random forest, hyperparameters such as the maximum depth and the minimum number of samples for splitting are tuned to balance the relationship between prediction accuracy and the generalization performance of the model.
[0018] Optionally, based on the selected connection metrics, the construction method of the prediction model also includes other statistical methods and machine learning methods, which will not be elaborated here.
[0019] When the applicant applies it specifically, the prediction effect based on the metabolic connectomics metrics is much higher than the traditional metabolicomics metrics modeling method. This is because the metabolic connectomics metrics themselves contain the information contained in the metabolic metrics, but at the same time utilize the interaction relationships between the metrics, which more conforms to the regulatory relationship changes of the in-vivo metabolic network under different disease states. Therefore, it is more meaningful for the classification and prediction of hepatic encephalopathy.
[0020] Optionally, the metabolite metabolome in the step S1 includes bile acids, long-chain fatty acids, short-chain fatty acids, neurotransmitters, and amino acid targeted metabolome.
[0021] According to another aspect of the present invention, there is also provided a hepatic encephalopathy recognition device based on metabolic connectomics for implementing the above-mentioned hepatic encephalopathy recognition method based on metabolic connectomics, including A data acquisition and cleaning unit for performing the step S1; A high-difference connection screening unit for high-difference connection screening; the high-difference connection screening includes the steps S21-S23; A key node screening unit; for key node screening, the key node screening includes the steps S31-S33; A benchmark pattern drawing unit for drawing a cold-hot connection pattern diagram; the drawing of the cold-hot connection pattern diagram includes the steps S41-S42; An individual quantification unit for individual connectomics quantification; the individual connectomics quantification includes the steps S51-S54; A pattern recognition unit for pattern recognition based on metabolic connectomics, and the pattern recognition based on metabolic connectomics includes the step S61.
[0022] According to another aspect of the present invention, there is also provided an electronic device, including: a processor and a memory; A computer-readable program executable by the processor is stored on the memory; When the processor executes the computer-readable program, the steps in the method as described above are implemented.
[0023] According to another aspect of the present invention, there is also provided a computer-readable storage medium storing one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the method as described above.
[0024] A method, device, equipment and medium for identifying encephalopathy in hepatic encephalopathy metabolic connectomics provided by the present invention not only considers the up-regulation and down-regulation changes of individual metabolomics indexes in the traditional sense, but also focuses on the key changes in the connection direction between metabolic indexes. By screening high-difference connections, key nodes in the metabolic network with frequently changed connections with other metabolites are obtained, and then the "cold" and "hot" connection modes of metabolites between the non-hepatic encephalopathy group and the hepatic encephalopathy group are obtained. Realize the mapping of individual connectomics quantitative data to "cold" and "hot" connection modes, achieve the individual classification goal, and thus solve the technical problems that the correlation coefficient matrix cannot be applied individually and the comprehensive modeling accuracy is poor in the related art. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 It is a schematic flowchart of a method for identifying encephalopathy in hepatic encephalopathy metabolic connectomics provided by an embodiment of the present invention; Figure 2 It is a schematic structural diagram of a device for identifying encephalopathy in hepatic encephalopathy metabolic connectomics provided by an embodiment of the present invention; Figure 3 It is a structural block diagram of a terminal that can implement the method for identifying encephalopathy in hepatic encephalopathy metabolic connectomics according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.
[0027] Embodiment 1 According to an embodiment of the present invention, there is provided a method for identifying encephalopathy in hepatic encephalopathy metabolic connectomics, in combination with Figure 1 , the method includes: Step S1: Obtain a metabolite dataset and perform logarithmic transformation on the metabolite index data in the metabolite dataset.
[0028] In step S1, the original metabolomics dataset can be read from a database or a file system. These data usually include the concentration values of various metabolites in different samples (patients with hepatic encephalopathy and control groups). The logarithmic transformation can be achieved by calling a logarithmic transformation algorithm to traverse the entire metabolite dataset. The transformed data is stored back into a data structure in memory or saved as an intermediate result file for use in subsequent steps.
[0029] Each metabolome includes bile acids, long-chain fatty acids, short-chain fatty acids, neurotransmitters, and amino acid targeted metabolomes.
[0030] Step S2: High-difference connection screening; the high-difference connection screening includes: Step S21, calculate the Pearson correlation coefficient between each pair of all transformed metabolite indices, separately for hepatic encephalopathy and the control group.
[0031] In step S21, as an example, for the metabolomics dataset that has undergone logarithmic transformation, the Pearson function can be called. For each metabolite, calculate the Pearson correlation coefficient between all pairs of metabolites within the hepatic encephalopathy group and the control group respectively, and store the calculated correlation coefficients as two matrices, one for the hepatic encephalopathy group and the other for the control group.
[0032] Step S22, compare the correlation coefficients between the two groups. If the positive and negative signs are opposite, it is confirmed as a directional difference. If the signs are the same and the difference in the correlation coefficients is greater than or equal to a preset threshold, it is confirmed as a strong difference, otherwise it is a weak difference. This setting can effectively capture the significant connection changes between metabolites in the disease state.
[0033] In step S22, as an example, if the positive and negative signs are opposite, it is defined as a "directional difference". If the signs are the same and the difference in the correlation coefficients ≥ 0.6, it is defined as a "strong difference", otherwise it is a weak difference (the 0.6 threshold is an empirical suggestion and can be adjusted).
[0034] Step S23, output the metabolite pairs with directional differences and strong differences and confirm them as high-difference connections.
[0035] In step S2, first calculate the correlation coefficient between each pair of indices, then compare the differences in the correlation coefficients between different disease state groups, and screen for positive and negative differences or high-value differences, which are defined as high-difference connections.
[0036] Step S3: Key node screening; the key node screening includes: Step S31: Decompose the high - difference connections, and split each connection into two nodes.
[0037] In step S31, the high - difference connections obtained in step S23 can be traversed, and then each high - difference connection is decomposed into two nodes.
[0038] Step S32: The nodes have no direction, and mix the decomposed nodes. All the split nodes can be put into a set, or other appropriate data structures such as lists can be used.
[0039] Step S33: Count the frequencies of different metabolites as nodes, and extract the top n nodes with the highest frequencies as key nodes.
[0040] In step S33, using a suitable data structure, count the frequency of each node. For example, in Python, an empty dictionary can be initialized, and then the node list is traversed. For each node, increase its count value in the dictionary. Sort the nodes according to the frequencies, and select the top n nodes with the highest frequencies as "key nodes" (n is an integer greater than 2). It should be noted here that according to the requirements, the "key nodes" must have high - difference connections with more than 50% of other metabolites. Therefore, in practical applications, it may be necessary to further verify whether the selected nodes meet this condition.
[0041] In step S3, decompose the high - difference connections into 2 connection nodes item by item, then mix and count the connection nodes, and finally define the high - frequency (with high - difference connections with > 50% metabolites) as key nodes.
[0042] Step S4: Benchmark pattern drawing; the benchmark pattern drawing includes: Step S41: Select the key nodes, and repeat step S21 to calculate the correlation matrices for the hepatic encephalopathy group and the control group respectively. The specific calculation method can refer to the introduction of step S21 above and will not be elaborated here.
[0043] Step S42: Draw a heatmap based on the correlation matrix, obtaining two opposite results of cold and heat, and thus obtaining two different connection patterns for hepatic encephalopathy and the control, which are used as the heat - related benchmark matrix and the cold - related benchmark matrix respectively.
[0044] In step S42, when drawing the heatmap, by comparing the differences in correlation coefficients between the two groups, two distinct connection patterns of heat and cold are identified. The heat pattern usually indicates a strong positive or negative correlation, while the cold pattern represents a weak correlation or the lack of a clear connection. This distinction is jointly ensured by the high - difference connection and key node screening in the previous steps.
[0045] In step S4, a chord diagram is drawn for the key nodes under the disease grouping, and the respective correlation matrices are defined as the heat / heat reference matrices.
[0046] Since individual data cannot be used to calculate the connection pattern, which affects the individualized application of metabolic connectomics, to overcome this problem, the following method is further adopted to solve it.
[0047] Step S5: Individual connectomics quantification; the individual connectomics quantification includes: Step S51: For hepatic encephalopathy and the control group respectively, calculate the percentile of the key nodes contained in the pattern reference item by item, and generate the hepatic encephalopathy position samples (position samples 1 - 3) and the control position samples (position samples 4 - 6).
[0048] In step S51, as an example, for each group (hepatic encephalopathy and control), calculate the P25, P50, and P75 percentiles of each key node based on the key nodes. According to these percentile values, generate position samples 1 - 3 for the hepatic encephalopathy group, which are P25, P50, and P75 respectively; similarly, generate position samples 4 - 6 for the control group.
[0049] Step S52: Merge the individual data with the hepatic encephalopathy position samples (position samples 1 - 3), and repeat step S21, that is, calculate the Pearson correlation coefficient between all key nodes to obtain the heat-related mapping matrix; Step S53: Merge the individual data with the control position samples (position samples 4 - 6), and repeat step S21 to obtain the cold-related mapping matrix; Step S54: Based on steps S52 and S53, record the correlation coefficients between the key nodes in pairs as the heat connection index and the cold connection index respectively.
[0050] In step S5, calculate the position parameters P25, P50, and P75 under the grouping, mix the new individual data with the position parameters of different groups respectively to obtain the correlation matrix, which is defined as the projection matrix, and finally calculate 2C(n,2) connectomics indicators based on the key node correlation coefficients.
[0051] In step S5, based on the two correlation coefficient matrices, the correlation coefficient values between every two key nodes are extracted respectively, denoted as the hot connection index and the cold connection index. Since each key node forms a relationship with other nodes, the total number of metabolic connectomics indicators will reach the combination number C(n, 2), where n is the number of key nodes. In contrast, traditional metabolomics usually only focuses on fewer indicators (r < n). In step S5, a data structure can be created to store these indicators for subsequent analysis or modeling. Through the design of the above technical solution, a breakthrough in individualized application is achieved, enabling the data of a single individual to participate in the connectomics analysis, thus supporting more accurate classification of the disease state of hepatic encephalopathy and the development of personalized medical strategies. This process not only improves the flexibility and accuracy of data analysis but also provides a solid foundation for further research.
[0052] Step S6: Pattern recognition based on metabolic connectomics; the pattern recognition based on metabolic connectomics includes: Step S61, draw graphs according to the hot correlation mapping matrix and the cold correlation mapping matrix to visualize the pattern similarity with the hot correlation benchmark matrix and the cold correlation benchmark matrix.
[0053] In step S61, as an example, visualization tools (such as Seaborn or Matplotlib) can be used to draw heatmaps to compare the "hot correlation mapping matrix" of an individual with the "hot correlation benchmark matrix", and the "cold correlation mapping matrix" with the "cold correlation benchmark matrix". These heatmaps can help visually evaluate the approximation of the individual's pattern to that of the hepatic encephalopathy group or the control group. For the correlation coefficients close to 1 or -1 in the mapping graph, it indicates that the individual pattern is more inclined to a certain category (hepatic encephalopathy or control).
[0054] Step S62, for cases that are not easy to distinguish, calculate the similarity between the hot correlation mapping matrix and the cold correlation mapping matrix and the hot correlation benchmark matrix and the cold correlation benchmark matrix respectively, and then complete pattern recognition, that is, disease state classification.
[0055] In step S62, as an example, for those cases that are difficult to judge visually, calculate the similarity between the "hot correlation mapping matrix" and the "hot correlation benchmark matrix", and the "cold correlation mapping matrix" and the "cold correlation benchmark matrix". Pearson correlation coefficient or other similarity measurement methods (such as cosine similarity) can be used for quantitative comparison. According to the similarity score, determine which category the individual is more inclined to belong to. Empirically, the "hot correlation mapping matrix" and the "cold correlation mapping matrix" of the same individual do not differ much, and a simple method is to analyze based on the "hot correlation mapping matrix".
[0056] As an example, to calculate the similarity between the heat-related mapping matrix, cold-related mapping matrix and the heat-related reference matrix, cold-related reference matrix, the Frobenius norm distance can be used. The smaller the distance, the more similar. The calculation formula is: where \(C\) is the mapping matrix to be determined, and \(A\) is the reference matrix.
[0057] In step S6, perform a similarity test on the mapping matrix and the reference matrix, objectively evaluate the group to which the individual belongs, and draw a heat map or chord diagram based on the above matrices to visually identify the metabolic pattern to which the individual belongs.
[0058] Step S7, prediction based on metabolomic connectomics indicators; the prediction based on metabolomic connectomics indicators includes: Step S71, screen the heat connection indicators and cold connection indicators.
[0059] In step S71, in addition to the matrix similarity method (global method), different statistical and machine learning methods can also be used to screen the connection indicators, including but not limited to methods such as AUC value evaluation, Lasso regression, elastic net, random forest variable importance, and SHAP value. In the analysis of this embodiment, it is considered to evaluate each connection indicator. Among them, 4 "heat connection indicators" have an AUC > 0.9 for differentiating hepatic encephalopathy and meet the biological significance, and 20 connection indicators have an AUC > 0.8. Both the values and the quantity are much higher than the metabolomics indicators.
[0060] Step S72, based on the screened connection indicators, construct a prediction model. And it can be used in combination with traditional omics indicators (the information quantification angles of the two are different, so the correlation is not high). The modeling can be based on statistical methods, machine learning, deep learning, and others.
[0061] As an example, the construction of the prediction model can adopt the random forest method.
[0062] In the random forest method, using the screened connection indicators as the variable set and whether it is hepatic encephalopathy as the classification result, combine to generate a data set. The modeling method is to randomly select samples from the data set with replacement for training to construct decision trees. Each time a split occurs in the decision tree, randomly select a part of the variable set to avoid overfitting of the model. Integrate multiple decision trees to form a random forest, and determine the final prediction result based on the highest frequency classification in all decision trees. During the construction of the random forest, tune hyperparameters such as the maximum depth and the minimum number of samples for splitting to balance the relationship between prediction accuracy and model generalization performance.
[0063] Step S8, disease state judgment based on the heat-cold related matrix and key connection indicators. The steps for disease judgment are: Step S81: For each sample to be detected, obtain the heat-related mapping matrix and cold-related mapping matrix generated in steps S52 and S53 respectively, and the key connection metrics extracted in step S54. Step S82: Data difference calculation. For the heat connection metrics: Calculate the difference between the correlation coefficients between each pair of key nodes in the sample and the corresponding values in the pre-established heat-related benchmark matrix, and take the average or weighted average of all the differences as the heat metric deviation; for the cold connection metrics: Calculate the difference between the correlation coefficients between each pair of key nodes in the sample and the corresponding values in the cold-related benchmark matrix, and also obtain the cold metric deviation.
[0064] Step S83: Comprehensive score generation. After normalizing the heat metric deviation and the cold metric deviation respectively, calculate a comprehensive diagnosis score based on the preset weights (the weights can be set according to the results of statistical analysis and clinical verification).
[0065] Step S84: Disease state judgment. Compare the calculated comprehensive score with a pre-set threshold. If the score is greater than or equal to the threshold, judge that the sample is in the state of hepatic encephalopathy; otherwise, judge it as not in the state of hepatic encephalopathy. The setting of the threshold can be based on the ROC curve analysis and statistical optimization of a large amount of known case data, so as to ensure the accuracy and robustness of this judgment method in practical applications.
[0066] Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases, the former is a better implementation method. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc), and includes several instructions for causing a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present application.
[0067] Embodiment 2 According to an embodiment of the present invention, in combination with Figure 2 , a device for identifying hepatic encephalopathy based on metabolic connectomics is provided, which is used to implement the above-mentioned analysis method of hepatic encephalopathy metabolic connectomics, and includes: A data acquisition and cleaning unit for executing the step S1; A high-difference connection screening unit for high-difference connection screening; the high-difference connection screening includes the steps S21 - S23; A key node screening unit for key node screening, and the key node screening includes the steps S31 - S33; A reference mode drawing unit for reference mode drawing; the reference mode drawing includes the steps S41 - S42; An individual quantification unit for individual connectomics quantification; the individual connectomics quantification includes the steps S51 - S54.
[0068] A pattern recognition unit for pattern recognition based on connectomics; the pattern recognition based on connectomics includes the above - mentioned steps S61 - S62.
[0069] A direct classification unit for direct classification based on connectomics; the direct classification based on connectomics includes the above - mentioned steps S71 - S72.
[0070] A disease state judgment unit for judging the disease state of an individual; the disease state judgment based on individual connectomics quantification includes S81 - S84.
[0071] Optionally, the specific examples in this embodiment can refer to the examples described in the above - mentioned embodiment, and will not be elaborated herein.
[0072] It should be noted here that the examples and application scenarios implemented by the above - mentioned modules and the corresponding steps are the same, but are not limited to the content disclosed in the above - mentioned embodiment. It should be noted that the above - mentioned modules, as part of the device, can run in the corresponding hardware environment, can be implemented by software, or can be implemented by hardware, where the hardware environment includes a network environment.
[0073] Figure 3 is a structural block diagram of a terminal according to an embodiment of the present application, as Figure 3 shown. The terminal may include: one or more (only one is shown) processors 101, a memory 103, and a transmission device 105, as Figure 3 shown. The terminal may further include an input / output device 107.
[0074] Among them, the memory 103 can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and devices in the embodiments of the present application. The processor 101 executes various functional applications and data processing by running the software programs and modules stored in the memory 103, that is, implements the above - mentioned method. The memory 103 may include a high - speed random access memory, and may further include a non - volatile memory, such as one or more magnetic storage devices, flash memory, or other non - volatile solid - state memories. In some instances, the memory 103 may further include a memory remotely set relative to the processor 101, and these remote memories can be connected to the terminal through a network. Examples of the above - mentioned network include, but are not limited to, the Internet, an enterprise internal network, a local area network, a mobile communication network, and combinations thereof.
[0075] The above-mentioned transmission device 105 is used to receive or send data via a network, and can also be used for data transmission between a processor and a memory. Specific examples of the above-mentioned network can include a wired network and a wireless network. In one example, the transmission device 105 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers through a network cable, so as to communicate with the Internet or a local area network. In one example, the transmission device 105 is a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0076] Specifically, the memory 103 is used to store application programs.
[0077] The processor 101 can call the application programs stored in the memory 103 through the transmission device 105 to execute the steps in the above method. Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments, and will not be elaborated herein.
[0078] Those of ordinary skill in the art can understand that the structure of the above terminal is only illustrative. The terminal can be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, and a Mobile Internet Devices (MID), a PAD and other terminal devices. Figure 3 It does not limit the structure of the above electronic device. For example, the terminal may further include more or fewer components (such as a network interface, a display device, etc.) than those shown Figure 3 in the figure, or have a different configuration from that shown Figure 3 in the figure.
[0079] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program. The program can be stored in a computer-readable storage medium. The storage medium can include: a flash drive, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disc, etc.
[0080] The embodiment of the present application also provides a storage medium. Optionally, in this embodiment, the above storage medium can be used to execute the program code of the above method.
[0081] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.
Claims
1. A method for identifying hepatic encephalopathy based on metabolic connectomics, characterized in that, Including: Step S1: Obtain a metabolite dataset and perform logarithmic transformation on the metabolite data in the metabolite dataset; Step S2: High-difference connection screening; The high-difference connection screening includes: Step S21: Calculate the Pearson correlation coefficient between every two of all the transformed metabolite indicators, respectively for the hepatic encephalopathy group and the control group; Step S22: Compare the correlation coefficients between the two groups. If the positive and negative signs are opposite, it is confirmed as a directional difference. If the signs are the same and the difference in the correlation coefficients is greater than or equal to a preset threshold, it is confirmed as a strong difference. Otherwise, it is a weak difference; Step S23: Output the metabolite pairs with the directional difference and the strong difference and confirm them as high-difference connections; Step S3: Key node screening; The key node screening includes: Step S31: Decompose the high-difference connections, and split each connection into two nodes; Step S32: Mix the decomposed nodes; Step S33: Count the frequencies of different metabolites as nodes, and extract the top n nodes with the highest frequencies as key nodes; Step S4: Benchmark pattern drawing; The benchmark pattern drawing includes: Step S41: Select the key nodes and repeat Step S21 to calculate the correlation matrices for the hepatic encephalopathy group and the control group respectively; Step S42: Draw a heat map based on the correlation matrices to obtain two different connection patterns for hepatic encephalopathy and the control, which are used as the heat-related benchmark matrix and the cold-related benchmark matrix respectively; Step S5: Individual connectomics quantification; The individual connectomics quantification includes: Step S51: Calculate the percentiles of the key nodes contained in the pattern benchmarks for the hepatic encephalopathy group and the control group respectively to generate the hepatic encephalopathy position sample and the control position sample; Step S52: Combine the individual data with the hepatic encephalopathy position sample and repeat Step S21 to obtain the heat-related mapping matrix; Step S53: Combine the individual data with the control position sample and repeat Step S21 to obtain the cold-related mapping matrix; Step S54: Based on Step S52 and Step S53, record the correlation coefficients between every two of the key nodes as the heat connection index and the cold connection index respectively; Step S6: Identification of hepatic encephalopathy based on metabolic connectomics; The identification of hepatic encephalopathy based on metabolic connectomics includes: Step S61: Draw a graph according to the heat-related mapping matrix and the cold-related mapping matrix to visualize the pattern similarity with the heat-related benchmark matrix and the cold-related benchmark matrix.
2. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 1, wherein, Step S6 further includes: Step S62: For cases that are not easily distinguishable, calculate the similarities between the heat-related mapping matrix, the cold-related mapping matrix and the heat-related benchmark matrix, the cold-related benchmark matrix respectively, and then complete pattern recognition, that is, disease state classification.
3. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 2, wherein To calculate the similarities between the heat-related mapping matrix, the cold-related mapping matrix and the heat-related benchmark matrix, the cold-related benchmark matrix, the Frobenius norm distance can be used. The smaller the distance, the more similar. The calculation formula is: where C is the mapping matrix to be determined and A is the benchmark matrix.
4. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 1, characterized in that, The method further includes: Step S7: Prediction based on metabolic connectomics indicators; The prediction based on metabolic connectomics indicators includes: Step S71: Screen the hot connection metrics and cold connection metrics; Step S72: Based on the screened connection metrics, construct a prediction model.
5. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 4, characterized in that The screening method for screening the hot connection metrics and cold connection metrics includes AUC, Lasso, elastic net, random forest variable importance, and SHAP.
6. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 1, wherein The construction of the prediction model includes the random forest method. Using the screened connection metrics as the variable set and whether it is hepatic encephalopathy as the classification result, a dataset is merged and generated. Samples are randomly selected from the dataset with replacement for training to construct decision trees. When splitting each time in the decision tree, a partial variable set is randomly selected to avoid overfitting of the model. Multiple decision trees are integrated to form a random forest, and the final prediction result is determined by the highest frequency classification among all decision trees.
7. The method for identifying hepatic encephalopathy based on metabolic connectomics according to claim 1, wherein During the construction of the random forest, hyperparameters such as the maximum depth and the minimum number of samples for splitting are tuned to balance the relationship between prediction accuracy and the generalization performance of the model.
8. A hepatic encephalopathy recognition device based on metabolic connectomics, which is used to implement the hepatic encephalopathy recognition method based on metabolic connectomics according to claim 1, and is characterized in that, Including A data acquisition and cleaning unit for performing the said Step S1; A high-difference connection screening unit for high-difference connection screening; the high-difference connection screening includes the said Steps S21 - S23; A key node screening unit; For key node screening, the key node screening includes the said Steps S31 - S33; A reference pattern drawing unit for drawing hot and cold connection pattern diagrams; The drawing of the hot and cold connection pattern diagrams includes the said Steps S41 - S42; An individual quantification unit for individual connectomics quantification; The individual connectomics quantification includes the said Steps S51 - S54; A pattern recognition unit for pattern recognition based on metabolic connectomics, and the pattern recognition based on metabolic connectomics includes the said Step S61.
9. An electronic device, characterized in that, Including: A processor and a memory; The memory stores a computer-readable program executable by the processor; When the processor executes the computer-readable program, it implements the steps in the method according to any one of claims 1 - 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the method according to any one of claims 1 - 7.