Hepatitis b diagnosis index extraction method based on blood detection and complex network robustness

Abstract

The application discloses a hepatitis B diagnosis index extraction method based on blood detection and complex network resilience, first, collecting blood routine, hepatitis B five items, liver function examination, blood coagulation four items, hepatitis B virus quantification, alpha-fetal protein data of patients with chronic hepatitis B, cirrhosis and hepatocellular carcinoma after onset without treatment; then extracting 25 key indexes, reassigning each index according to the liver disease diagnosis standard treatment guide and Child-Pugh score; next, calculating the correlation between indexes and setting a threshold, respectively constructing the index correlation degree network of chronic hepatitis B, cirrhosis and hepatocellular carcinoma; calculating the average degree of the network; processing the data again and calculating the mean value, fitting with neural network dynamics to obtain the fitting error; obtaining the diagnosis interval corresponding to chronic hepatitis B, cirrhosis and hepatocellular carcinoma, which is used as the diagnosis index of hepatitis B patients. The method is simple and efficient in modeling process, and is beneficial to improving the utilization rate of medical resources.

Description

Technical Field

[0001] This invention belongs to the field of biomedical electro-technical technology, specifically relating to a method for extracting diagnostic indicators for hepatitis B. Background Technology

[0002] Early assessments of liver fibrosis, particularly in resource-scarce regions, used serological markers for hepatitis B, serum HBV-DNA quantification, serum alanine aminotransferase (ALT) concentration, and the aspartate aminotransferase (AST) to platelet (PLT) ratio. However, these methods are insufficient for a comprehensive evaluation of liver function in chronic hepatitis B (CHB), cirrhosis (LC), and hepatocellular carcinoma (HCC). Due to limited sensitivity, most international recommendations no longer recommend serum alpha-fetoprotein (AFP) for diagnosing HCC (in fact, in some guidelines, it is used in conjunction with radiographic features), but it remains widely used in West Africa. Meanwhile, common imaging methods also have drawbacks. While ultrasound can dynamically reflect the blood supply characteristics of lesions in real time, it has limitations because abdominal organs and gases can obstruct ultrasound diagnosis. Furthermore, the sonographer's skill and personal judgment can affect the accuracy of ultrasound examinations. CT can accurately assess the relationship between lesions and surrounding structures, but it involves radiation and has low soft tissue resolution. Due to radiation absorption, repeated CT scans are not recommended for patients within a short period. Magnetic resonance imaging (MRI) scans offer multi-parameter, multi-directional imaging capabilities, and the use of hepatocellular-specific contrast agents can improve the detection rate and diagnostic accuracy of small hepatocellular carcinomas. However, a drawback of MRI is its relatively high requirement for patient compliance, as patients must remain in a fixed position for a certain period. Another common diagnostic tool for liver disease—liver biopsy—can lead to complications, including pain and bruising at the biopsy site, prolonged bleeding, infection near the biopsy site, and accidental damage to another organ, especially for patients with advanced liver disease or cancer who are unsuitable for liver biopsy. In summary, all of the above methods can potentially impose financial burdens or physical harm on patients. Since all patient visits require blood tests, integrating blood test results for a comprehensive assessment of liver function is feasible. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention provides a method for extracting hepatitis B diagnostic indicators based on blood testing and the resilience of complex networks. First, it collects blood routine, hepatitis B five-item, liver function test, coagulation four-item, hepatitis B virus quantification, and alpha-fetoprotein data from patients with chronic hepatitis B, cirrhosis, and hepatocellular carcinoma before treatment. Then, it extracts 25 key indicators and re-assigns values ​​to each indicator according to liver disease diagnostic criteria, treatment guidelines, and the Child-Pugh score. Next, it calculates the correlation between indicators and sets thresholds, constructing correlation networks for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma indicators respectively; it calculates the network mean degree; it processes the data again and calculates the mean, fitting it with neural network dynamics to obtain the fitting error; and it determines the diagnostic intervals for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma, using these as diagnostic indicators for hepatitis B patients. This invention's method has a simple and efficient modeling process, which is beneficial for improving the utilization rate of medical resources.

[0004] The technical solution adopted by this invention to solve its technical problem includes the following steps:

[0005] Step 1: Collect blood routine, hepatitis B five items, liver function test, coagulation four items, hepatitis B virus quantitative and alpha-fetoprotein data of patients with chronic hepatitis B, cirrhosis and hepatocellular carcinoma before treatment after onset;

[0006] Step 2: Based on the data collected in Step 1, extract 25 key indicators, and regroup and reassign values ​​to each indicator according to the diagnostic criteria and treatment guidelines for liver diseases and the Child-Pugh score;

[0007] Step 2-1: Based on the data collected in Step 1, extract the following 25 indicators: red blood cell count (RBC), white blood cell count (WBC), platelet count (PLT), hemoglobin (HGB), hepatitis B surface antigen (HBsAg), hepatitis B surface antibody (HBsAb), hepatitis B e antigen (HBeAg), hepatitis B e antibody (HBeAb), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), total cholesterol (CHOL), total bilirubin (TBILI), direct bilirubin (DBILI), indirect bilirubin (IBILI), total protein (TP), albumin (ALB), globulin (GLB), prothrombin time (PT), fibrinogen (FIB), activated partial thromboplastin time (APTT), plasma thrombin time (TT), HBV-DNA quantification, and alpha-fetoprotein (AFP).

[0008] Step 2-2: Reassign values ​​to each indicator group according to the diagnostic criteria and treatment guidelines for liver diseases and the Child-Pugh score;

[0009] Step 3: Calculate the correlation between the indicators and set thresholds for the data obtained in Step 2, and construct correlation networks for indicators of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma respectively.

[0010] Step 4: Calculate the average network degree for the three networks obtained in Step 3: chronic hepatitis B, cirrhosis, and hepatocellular carcinoma.

[0011] Step 5: Based on the network obtained in Step 3 and the average network degree obtained in Step 4, process the data again according to the formula and calculate the mean, fit it with the neural network dynamics, and obtain the fitting error;

[0012] Step 6: Using the mean and fitting error obtained in Step 5, calculate the diagnostic intervals for the three diseases: chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. These intervals will serve as diagnostic indicators for hepatitis B patients.

[0013] Furthermore, in step 1, the patients must have a history of hepatitis B, and the collected data must be blood tests performed when the patient first developed the disease and was not treated. All patients must not have ascites or hepatic encephalopathy, and the Child-Pugh score of cirrhotic patients must be A or B.

[0014] Furthermore, in step 2-2, each indicator group is re-assigned a value of 2 or 4.

[0015] Furthermore, step 3 is specifically as follows:

[0016] Calculate the correlation coefficient between the indicators using the Pearson correlation coefficient formula:

[0017]

[0018] Where cov(x) i x j ) are two indicators x i and x j covariance, and These are two indicators x i and x j The standard deviation of the network; therefore, the adjacency matrix of the network is defined as:

[0019]

[0020] This allows for the establishment of a correlation network of indicators in cases of chronic hepatitis B, cirrhosis, or hepatocellular carcinoma.

[0021] Furthermore, step 4 is specifically as follows:

[0022] First, remove isolated nodes from the network; in the network, the degree of a node is the total number of edges connecting a node i to all other nodes in the network. Therefore, the average degree of the network <s>The calculation formula is:

[0023]

[0024] Where N is the number of network nodes, i.e. the number of indicators; and These are the positive and negative adjacency matrices of the network, respectively, and they satisfy:

[0025]

[0026] Furthermore, step 5 is specifically as follows:

[0027] Step 5-1: Obtain the network from Step 3 and the average network degree from Step 4, according to the formula:

[0028]

[0029] The data is processed again, and then the mean formula is applied:

[0030]

[0031] Calculate the mean of the processed indicators;

[0032] Step 5-2: Assuming a positive and negative adjustment relationship exists between each pair of indicators, a neural network dynamics equation is used to simulate the dynamic changes between the indicators:

[0033]

[0034] Among them I i It is the change of node i itself, R i It is the reciprocal of the proportion by which node i returns to its normal range, and N is the number of non-isolated nodes. and The excitation and inhibition strengths of node i are calculated as follows:

[0035]

[0036] Where P ij and Q ij These are the positive and negative correlation coefficient matrices, respectively;

[0037] Step 5-3: Set the value of index j to x i Replace with node average<x′> Then the original differential equation (1) is approximately:

[0038]

[0039] in and These are the positive and negative node degrees of index i, respectively;

[0040] Because the hyperbolic tangent function tanh(x) can be expressed as:

[0041]

[0042] Therefore, equation (2) can be written as:

[0043]

[0044] make

[0045]

[0046] When the system reaches a steady state, equation (3) must satisfy:

[0047]

[0048] Step 5-4: According to formula (4), the steady-state behavior value of index i is:

[0049]

[0050] Because of the limit:

[0051]

[0052] Then formula (5) can be approximated as:

[0053]

[0054] Let p represent the proportion of negatively correlated edges between nodes:

[0055]

[0056] Based on the formula for the mean, and according to formulas (2), (7), (8) and the network average degree obtained in step 4. <s>Decoupling formula (1) results in:

[0057] <x>=I i R i +R i ·J′1· <s>·(1-p) (9)

[0058] in It is the excitation intensity of the network;

[0059] Therefore, setting n=2 and a=-1, the unknown parameter I under the conditions of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma can be obtained by solving equations (6) and (9) simultaneously. i and R i Finally, the fitting error ξ is calculated under each of the three states. k Where k = 1, 2, 3 correspond to chronic hepatitis B, cirrhosis, and hepatocellular carcinoma, respectively.

[0060] Preferably, the method for determining the diagnostic intervals corresponding to the three diseases—chronic hepatitis B, cirrhosis, and hepatocellular carcinoma—in step 6 is as follows:

[0061] Mean values ​​from chronic hepatitis B, cirrhosis, and hepatocellular carcinoma states <x′ k >and fitting error ξ k Calculate the diagnostic interval according to the formula:

[0062]

[0063] Preferably, the method of the present invention can be used for the diagnosis of hepatitis B-related diseases.

[0064] The beneficial effects of this invention are as follows:

[0065] The proposed method for extracting hepatitis B diagnostic indicators based on blood tests and the resilience of complex networks requires only blood tests performed on patients with chronic hepatitis B, cirrhosis, or hepatocellular carcinoma before treatment. It extracts 25 indicators, constructs a network based on the correlations between these indicators, and analyzes its structural features. Simultaneously, it fits the network with neural network dynamics to derive a comprehensive evaluation index for hepatitis B liver function. The modeling process is simple and efficient. Attached Figure Description

[0066] Figure 1 This is a flowchart of the method of the present invention.

[0067] Figure 2 This is a network diagram illustrating the chronic hepatitis B state in an embodiment of the present invention.

[0068] Figure 3 This is a network diagram illustrating the state of liver cirrhosis in an embodiment of the present invention.

[0069] Figure 4 This is a schematic diagram of the network in the hepatocellular carcinoma state according to an embodiment of the present invention. Detailed Implementation

[0070] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0071] To avoid redundant testing in existing diagnostic methods and reduce the stress and harm to patients caused by high costs during the examination process, this invention proposes a method for extracting hepatitis B diagnostic indicators based on blood testing and the resilience of complex networks. This method applies complex network concepts to the diagnosis of patients with chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. By collecting and processing routine blood tests in untreated cases, calculating their correlation coefficients, constructing a relationship network between indicators, analyzing the network structure characteristics under different disease conditions, and determining the corresponding network average degree, and combining this with neural network dynamics, diagnostic indicators for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma are obtained.

[0072] A method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience, specifically:

[0073] Step 1: Collect blood routine, hepatitis B five items, liver function test, coagulation four items, hepatitis B virus quantitative and alpha-fetoprotein data of patients with chronic hepatitis B, cirrhosis and hepatocellular carcinoma before treatment after onset;

[0074] Step 2: Based on the data collected in Step 1, extract 25 key indicators, and regroup and reassign values ​​to each indicator according to the diagnostic criteria and treatment guidelines for liver diseases and the Child-Pugh score;

[0075] Step 2-1: Based on the blood test data collected in Step 1, extract 25 indicators including red blood cell count (RBC), white blood cell count (WBC), platelet count (PLT), hemoglobin (HGB), hepatitis B surface antigen (HBsAg), hepatitis B surface antibody (HBsAb), hepatitis B e antigen (HBeAg), hepatitis B e antibody (HBeAb), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), total cholesterol (CHOL), total bilirubin (TBILI), direct bilirubin (DBILI), indirect bilirubin (IBILI), total protein (TP), albumin (ALB), globulin (GLB), prothrombin time (PT), fibrinogen (FIB), activated partial thromboplastin time (APTT), plasma thrombin time (TT), HBV-DNA quantification, and alpha-fetoprotein (AFP).

[0076] Step 2-2: Reassign values ​​(2 or 4 values) to each indicator group according to the diagnostic criteria and treatment guidelines for liver disease and the Child-Pugh score.

[0077] Step 3: Calculate the correlation between the indicators and set thresholds for the data obtained in Step 2, and construct correlation networks for indicators of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma respectively.

[0078] Step 4: Analyze the structural characteristics of the three networks obtained in Step 3 (chronic hepatitis B, cirrhosis, and hepatocellular carcinoma) and calculate the average degree of the networks.

[0079] Step 5: Based on the network obtained in Step 3 and the average network degree obtained in Step 4, process the data again according to the formula and calculate the mean, fit it with the neural network dynamics, and obtain the fitting error.

[0080] Step 6: Using the mean and fitting error obtained in Step 5, calculate the diagnostic intervals for the three diseases: chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. These intervals will serve as the diagnostic criteria for hepatitis B patients.

[0081] In step 1, the patient must have a history of hepatitis B, and the data collected must be the blood tests performed when the patient first developed the disease and was not treated. All patients must not have ascites or hepatic encephalopathy, and patients with cirrhosis must have a Child score of A or B.

[0082] In step 2, the data grouping must be reassigned according to the hepatitis B disease guidelines and the normal test range.

[0083] The threshold for the correlation coefficient in step 3 is usually set to 0.25. Specific implementation examples:

[0085] See Figures 1-4 This embodiment, based on blood testing and complex network resilience diagnostic technology for hepatitis B-related diseases, is applied to the diagnosis of hepatitis B-related diseases. The specific steps are as follows:

[0086] Step 1: Collect blood routine, hepatitis B five items, liver function test, coagulation four items, hepatitis B virus quantitative, and alpha-fetoprotein data from patients with chronic hepatitis B, cirrhosis, and hepatocellular carcinoma before treatment. The specific method is as follows:

[0087] In this embodiment, only six blood tests performed when the patient first seeks medical attention for the first time are considered: complete blood count, hepatitis B five markers, HBV DNA quantitative test, liver function test, coagulation four markers, and alpha-fetoprotein.

[0088] Step 2: Based on the blood test data of hepatitis B patients collected in Step 1, extract 25 key indicators. Reassign values ​​to each indicator according to the diagnostic criteria and treatment guidelines for liver diseases and the Child-Pugh score. The specific method is as follows:

[0089] Based on the blood test data collected in step 1, 25 indicators were extracted, including red blood cell count (RBC), white blood cell count (WBC), platelet count (PLT), hemoglobin (HGB), hepatitis B surface antigen (HBsAg), hepatitis B surface antibody (HBsAb), hepatitis B e antigen (HBeAg), hepatitis B e antibody (HBeAb), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), total cholesterol (CHOL), total bilirubin (TBILI), direct bilirubin (DBILI), indirect bilirubin (IBILI), total protein (TP), albumin (ALB), globulin (GLB), prothrombin time (PT), fibrinogen (FIB), activated partial thromboplastin time (APTT), plasma thrombin time (TT), HBV-DNA quantification, and alpha-fetoprotein (AFP). Each indicator was converted into a 2-value or 4-value according to the diagnostic criteria and treatment guidelines for liver disease and the Child-Pugh score, as shown in Table 1.

[0090] Table 1. Value Assignment Table for 25 Indicators

[0091]

[0092]

[0093] Step 3: For the reassigned data obtained in Step 2, calculate the correlation between indicators and set a threshold. Construct correlation networks for indicators under chronic hepatitis B, cirrhosis, or hepatocellular carcinoma, respectively. The specific methods are as follows:

[0094] Calculate the correlation coefficient between the indicators according to the Pearson correlation coefficient formula.

[0095]

[0096] Where cov(x) i x j ) are two indicators x i and x j covariance, and These are two indicators x i and x j The standard deviation of the network. Therefore, the adjacency matrix of the network is defined as...

[0097]

[0098] This allows for the establishment of a correlation network of indicators in cases of chronic hepatitis B, cirrhosis, or hepatocellular carcinoma.

[0099] Step 4: Analyze the structural characteristics of the three networks obtained in Step 3 (chronic hepatitis B, cirrhosis, and hepatocellular carcinoma) and calculate the average network degree. The specific method is as follows:

[0100] First, remove isolated nodes from the network. This is because the degree of a node is the total number of edges connecting it to other nodes in the network. Therefore, the average degree of the network <s>The calculation formula is:

[0101]

[0102] Where N is the number of network nodes, which is the number of metrics in this example. and These are the positive and negative adjacency matrices of the network, and they satisfy...

[0103]

[0104] Step 5: Based on the network obtained in Step 3 and the network average degree calculated in Step 4, process the data again according to the formula and calculate the mean. Fit the result with the neural network dynamics to obtain the fitting error. The specific method is as follows:

[0105] Based on the network obtained in step 3 and the average network degree calculated in step 4, according to the formula...

[0106]

[0107] The data is processed again, and then the mean formula is applied.

[0108]

[0109] Calculate the mean of the processed indicators.

[0110] Based on step 3, two different correlations were found between the detection indicators: positive or negative correlation. In fact, if there is a positive correlation between two indicators, an increase in one indicator will lead to an increase in the other; conversely, if there is a negative correlation between two indicators, an increase in one indicator will stimulate a decrease in the other. Assuming that there is a positive and negative moderating relationship between each pair of indicators, the dynamic changes between the indicators can be simulated using neural network dynamics equations.

[0111]

[0112] Among them I i It is the change of node i itself, R i It is the reciprocal of the proportion by which node i returns to its normal range, and N is the number of non-isolated nodes. and The excitation and inhibition strengths of node i are calculated as follows:

[0113]

[0114] Where P and Q are the positive and negative correlation coefficient matrices, respectively. Since I i and R i These are unknown parameters, requiring parameter estimation using data. The value of index j, x... j Replace with node average<x′> Then the original differential equation (1) can be approximated as

[0115]

[0116] in and These are the positive and negative node degrees of index i, respectively. Because the hyperbolic tangent function tanh(x) can be expressed as...

[0117]

[0118] Therefore, equation (2) can be written as

[0119]

[0120] make

[0121]

[0122] When the system reaches a steady state, equation (3) must satisfy...

[0123]

[0124] According to formula (4), the steady-state behavior value of index i is

[0125]

[0126] Because of the limit

[0127]

[0128] Then formula (5) can be approximated as

[0129]

[0130] Let p be the proportion of negatively correlated edges between nodes.

[0131]

[0132] Based on the formula for the mean, the average network degree can be obtained from formulas (2), (7), (8) and step 4. <s>Decouple equation (1) as follows

[0133] <x>=I i R i +R i ·J′1· <s>·(1-p) (9)

[0134] in It is the excitation intensity of the network. Therefore, setting n=2 and a=-1, the unknown parameter I under the conditions of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma can be obtained by solving equations (6) and (9) simultaneously. i and R i Finally, the fitting error ξ is calculated under each of the three states. k Where k = 1, 2, 3 correspond to chronic hepatitis B, cirrhosis, and hepatocellular carcinoma, respectively.

[0135] Step 6: Calculate the mean values ​​obtained in Step 5 for patients with chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. <x′ k >and fitting error ξ k The diagnostic intervals for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma are determined and used as diagnostic criteria for hepatitis B patients. The specific method is as follows:

[0136] The mean values ​​obtained in step 5 for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma states. <x′ k >and fitting error ξ k According to the formula

[0137]

[0138] Where k = 1, 2, 3 correspond to chronic hepatitis B, cirrhosis, and hepatocellular carcinoma, respectively. Therefore, the mean values ​​of the comprehensive assessment indicators in the chronic hepatitis B state correspond to the interval [0.4194, 0.5584]; the mean values ​​in the cirrhosis state correspond to the interval [0.4011, 0.4749]; and the mean values ​​in the cirrhosis state correspond to the interval (0.1353, 0.4292).

[0139] Therefore, the obtained intervals can be used as diagnostic criteria for chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. Although the average test results of a patient may fall into multiple intervals, the patient's condition can be determined by unique characteristics. For example, in the decompensated stage of cirrhosis, patients may experience hypersplenism, leading to a decrease in red blood cells, white blood cells, and platelets.

[0140] This invention can be used in the diagnosis of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. According to the method of this invention, the accuracy rates for identifying chronic hepatitis B, cirrhosis, and hepatocellular carcinoma are 84.21%, 86.67%, and 83.33%, respectively. The overall accuracy rate is 84.81%. However, according to surveys, the accuracy rate experienced by physicians without image-assisted diagnosis is only 56.97%. Therefore, the method of this invention can help physicians diagnose hepatitis B-related diseases based on only a small amount of blood sample.

[0141] This invention requires only a small blood sample to understand a patient's condition and provides a basis for treatment or referral based on liver function elasticity. For patients with severe malnutrition, prolonged clotting time, or large tumors, liver function can be assessed solely through blood tests, eliminating the need for liver biopsy. Globally, this will save at least $10.5 billion annually. Therefore, this method is beneficial for improving the utilization rate of medical resources and reducing the healthcare burden on individuals and nations.< / s> < / x> < / s> < / s> < / s> < / x> < / s> < / s>

Claims

1. A method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience, characterized in that, Includes the following steps: Step 1: Collect blood routine, hepatitis B five items, liver function test, coagulation four items, hepatitis B virus quantitative and alpha-fetoprotein data of patients with chronic hepatitis B, cirrhosis and hepatocellular carcinoma before treatment after onset; Step 2: Based on the data collected in Step 1, extract 25 key indicators, and regroup and revalue each indicator according to the diagnostic criteria and treatment guidelines for liver diseases and the Child-Pugh score; Step 3: Calculate the correlation between the indicators and set thresholds for the data obtained in Step 2, and construct correlation networks for indicators of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma respectively. Step 4: Calculate the average network degree for the three networks obtained in Step 3: chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. Step 4 is detailed below: First, remove isolated nodes from the network; in a network, a node... The degree of a node is the total number of edges connecting it to other nodes in the network. Therefore, the average degree of the network The calculation formula is: in This refers to the number of network nodes, i.e., the number of indicators. and These are the positive and negative adjacency matrices of the network, respectively, and they satisfy: Step 5: Based on the network obtained in Step 3 and the network average degree calculated in Step 4, process the data again according to the formula and calculate the mean, then fit it with the neural network dynamics to obtain the fitting error; Step 5 is detailed as follows: Step 5-1: Obtain the network from Step 3 and the average network degree from Step 4, according to the formula: The data is processed again, and then the mean formula is applied: Calculate the mean of the processed indicators; Step 5-2: Assuming a positive and negative adjustment relationship exists between each pair of indicators, a neural network dynamics equation is used to simulate the dynamic changes between the indicators: （1） in It is a node Its own changes It is a node The reciprocal of the proportion that returns to the normal range. Is the number of non-orphaned nodes? and It is a node The excitation and inhibition strengths are calculated as follows: （2） in P ij and Qij These are the positive and negative correlation coefficient matrices, respectively; Step 5-3: Set the indicators value Replace with node average Then the original differential equation (1) is approximately: （3） in and These are indicators Positive and negative node degree; Because the hyperbolic tangent function It can be represented as: ， Therefore, equation (2) can be written as: make （4） When the system reaches a steady state, formula (3) must satisfy: （5） Step 5-4: According to formula (4), the index The steady-state behavior value is: （6） Because of the limit: Then formula (5) can be approximated as: .（7） Let the proportion of negatively correlated edges between nodes be defined. Represented as: （8） Based on the formula for the mean, according to formulas (2), (7), (8) and the network average degree obtained in step 4. Decoupling formula (1) results in: （9） in It is the excitation intensity of the network; Therefore, setting and Solving equations (6) and (9) simultaneously yields the unknown parameters under chronic hepatitis B, cirrhosis, and hepatocellular carcinoma conditions. and Finally, the fitting error was calculated under each of the three conditions. ,in These correspond to the states of chronic hepatitis B, cirrhosis, and hepatocellular carcinoma, respectively. Step 6: Using the mean and fitting error obtained in Step 5, calculate the diagnostic intervals for the three diseases: chronic hepatitis B, cirrhosis, and hepatocellular carcinoma. These intervals will serve as diagnostic indicators for hepatitis B patients.

2. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 1, characterized in that, In step 1, the patients must have a history of hepatitis B, and the data collected must be blood tests performed when the patient first developed the disease and was not treated. All patients must not have ascites or hepatic encephalopathy, and the Child-Pugh score of patients with cirrhosis must be A or B.

3. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 2, characterized in that, Step 2 is described in detail below: Step 2-1: Based on the data collected in Step 1, extract the following 25 indicators: red blood cell count (RBC), white blood cell count (WBC), platelet count (PLT), hemoglobin (HGB), hepatitis B surface antigen (HBsAg), hepatitis B surface antibody (HBsAb), hepatitis B e antigen (HBeAg), hepatitis B e antibody (HBeAb), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), total cholesterol (CHOL), total bilirubin (TBILI), direct bilirubin (DBILI), indirect bilirubin (IBILI), total protein (TP), albumin (ALB), globulin (GLB), prothrombin time (PT), fibrinogen (FIB), activated partial thromboplastin time (APTT), plasma thrombin time (TT), HBV-DNA quantification, and alpha-fetoprotein (AFP). Step 2-2: Reassign values ​​to each indicator group according to the diagnostic criteria and treatment guidelines for liver disease and the Child-Pugh score.

4. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 3, characterized in that, In step 2-2, each indicator group is re-assigned a value of 2 or 4.

5. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 4, characterized in that, Step 3 is as follows: Calculate the correlation coefficient between the indicators using the Pearson correlation coefficient formula: in There are two indicators. and covariance, and These are two indicators. and The standard deviation of the network; therefore, the adjacency matrix of the network is defined as: This allows for the establishment of a correlation network of indicators in cases of chronic hepatitis B, cirrhosis, or hepatocellular carcinoma.

6. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 5, characterized in that, The method for determining the diagnostic intervals for the three diseases—chronic hepatitis B, cirrhosis, and hepatocellular carcinoma—in step 6 is as follows: Mean values ​​from chronic hepatitis B, cirrhosis, and hepatocellular carcinoma states and fitting error Calculate the diagnostic interval according to the formula: 。 7. The method for extracting hepatitis B diagnostic indicators based on blood testing and complex network resilience according to claim 1, characterized in that, It can be used for the diagnosis of hepatitis B-related diseases.

Images