A method for extracting bladder cancer biomarkers based on diagnostic models

Through the method based on diagnostic model, Bayesian network model is constructed and biomarkers are analyzed in association, which solves the time and resource consumption problems of traditional bladder cancer biomarker extraction methods, improves diagnostic accuracy and efficiency, and provides personalized treatment.

CN117831633BActive Publication Date: 2024-11-19JIANGSU HEFU BIOTECHNOLOGY CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202311724624.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-15
Publication Date
2024-11-19
Estimated Expiration
2043-12-15

AI Technical Summary

Technical Problem

Traditional bladder cancer biomarker extraction methods require a lot of time and resources, are uncomfortable to patients, and are complex and costly.

Method used

Using a diagnostic model-based method, by collecting patient medical record information and medical data, building a Bayesian network model, conducting biomarker correlation analysis, determining relevant markers, reducing invalid tests, and improving diagnostic accuracy.

Benefits of technology

It improves the accuracy and efficiency of bladder cancer diagnosis, reduces discomfort and resource consumption to patients, provides personalized treatment plans, reduces the number of test samples, and saves time and resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117831633B_ABST
    Figure CN117831633B_ABST
Patent Text Reader

Abstract

The present invention provides a method for extracting bladder cancer biomarkers based on a diagnostic model, which relates to the field of medical diagnosis and treatment technology. The method for extracting bladder cancer biomarkers based on a diagnostic model includes the following steps: collecting the patient's medical history information, and diagnosing bladder cancer for the patient based on a bladder cancer diagnostic model; collecting bladder cancer medical data, and performing correlation analysis between bladder cancer and biomarkers, and establishing a correlation database; determining the biomarkers associated with the patient's bladder cancer based on the patient's bladder cancer diagnosis results and the correlation database; and collecting the patient's test samples and extracting biomarkers based on the determined biomarkers. The present invention can determine possible biomarkers early, thereby reducing invalid or redundant tests, making the diagnostic process more efficient, and according to the determined biomarkers, only relevant samples need to be collected for testing, thereby reducing the number of samples that need to be tested, saving time and resources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of medical diagnosis and treatment technology, and in particular to a method for extracting bladder cancer biomarkers based on a diagnostic model. Background Art

[0002] Bladder cancer, a malignant tumor that originates from the lining or wall of the bladder, is a common cancer worldwide and is particularly common in men. The bladder is an important urinary system organ responsible for storing and excreting urine. Although our body has natural defense mechanisms to prevent excessive cell growth, in some cases these mechanisms may fail, causing cells to grow and divide without restriction, thus forming tumors. The formation of bladder cancer may be related to a variety of factors such as genetics, environment, and lifestyle. Among them, smoking, exposure to chemicals, and certain diseases (such as chronic bladder infections and bladder stones) may increase the risk of bladder cancer. Early symptoms of bladder cancer may include hematuria, frequent urination, and painful urination.

[0003] Biomarkers are quantifiable biological molecules that are usually present in body fluids or tissues and can be used to indicate the normal or abnormal state of biological processes, or to indicate the response to drug intervention. In the diagnosis of bladder cancer, biomarkers can help doctors more accurately judge the condition, guide treatment, and monitor the progression of the disease. At present, the traditional technology of biomarker extraction is to extract biomarkers from patients first, and then diagnose bladder cancer based on the judgment of biomarkers. This method requires the collection of a large number of test samples from patients, which not only requires a lot of time and resources, but also may cause unnecessary pain and discomfort to patients; in addition, some biomarkers may require special extraction and processing methods, further increasing the complexity and cost of the diagnostic process.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of this, the present invention provides a method for extracting bladder cancer biomarkers based on a diagnostic model to solve the problem that the traditional biomarker extraction technology mentioned above is to extract biomarkers from patients first, which not only requires a lot of time and resources, but also may cause unnecessary pain and discomfort to patients.

[0006] In order to solve the above problems, the specific technical solutions adopted by the present invention are as follows:

[0007] A method for extracting bladder cancer biomarkers based on a diagnostic model, the method comprising the following steps:

[0008] S1. Collect the patient's medical history information and diagnose bladder cancer for the patient based on the bladder cancer diagnosis model;

[0009] S2. Collect medical data on bladder cancer, conduct correlation analysis between bladder cancer and biomarkers, and establish a correlation database;

[0010] S3. determining a biomarker associated with bladder cancer of the patient based on the patient's bladder cancer diagnosis result and a related database;

[0011] S4. Based on the determined biomarkers, collect the patient's test samples and perform biomarker extraction.

[0012] Preferably, collecting the patient's medical history information and diagnosing bladder cancer for the patient based on the bladder cancer diagnosis model comprises the following steps:

[0013] S11. Collect the patient's medical history information, including the patient's personal medical history, family medical history, symptoms, and imaging examination results;

[0014] S12. Construction of bladder cancer diagnosis model based on structure optimization algorithm;

[0015] S13. Use the patient's medical history information as input and input it into the bladder cancer diagnosis model, and output the patient's bladder cancer diagnosis result.

[0016] Preferably, constructing a bladder cancer diagnosis model based on a structure optimization algorithm comprises the following steps:

[0017] S121, extracting diagnostic data of bladder patients and non-bladder patients from a hospital medical record system to form a data set;

[0018] S122, preprocessing the formed data set, and using the preprocessed data set as a training data set, the preprocessing includes data cleaning, abnormal data deletion and missing value supplementation;

[0019] S123, determining variable information related to bladder diagnosis according to medical literature, selecting the variable information from the training data set and using it as a Bayesian network node;

[0020] S124. Construct a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structural optimization algorithm.

[0021] Preferably, constructing a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structure optimization algorithm comprises the following steps:

[0022] S1241. For each pair of Bayesian network nodes, calculate the mutual information between them;

[0023] S1242, arranging the calculated mutual information in descending order, and adding edges in sequence to construct an initial structure of the Bayesian network;

[0024] S1243, performing a conditional independence test on the nodes without edges added in the initial structure of the Bayesian network, and determining whether there are implicit edges between the nodes without edges added, and if so, adding the implicit edges to the initial structure of the Bayesian network to obtain a secondary structure of the Bayesian network;

[0025] S1244, performing isolated node detection on the secondary structure of the Bayesian network, and calculating the mutual information between the isolated node and other nodes, connecting the node with the largest mutual information with the isolated node with an undirected edge, and obtaining the complete structure of the Bayesian network;

[0026] S1245. Perform structural search on the complete structure of the Bayesian network through a structural optimization algorithm, optimize the edges of the complete structure of the Bayesian network, obtain the optimal Bayesian network structure, and use it as a bladder cancer diagnosis model.

[0027] Preferably, the calculated mutual information is arranged in descending order, and edges are added in sequence. Constructing the initial structure of the Bayesian network includes the following steps:

[0028] S12421. Arrange the calculated mutual information in descending order to obtain a mutual information chain;

[0029] S12422. According to the obtained mutual information chain, node pairs are taken out in sequence, and it is determined whether there is a path between the node pairs;

[0030] S12423. If there is a path between the node pairs, an undirected edge is added between the node pairs, and the node pairs to which the undirected edge is added are deleted from the mutual information chain;

[0031] S12424. Repeat steps S12422-S12423 until the mutual information chain is empty.

[0032] Preferably, performing a structural search on the complete structure of the Bayesian network by a structural optimization algorithm, optimizing the edges of the complete structure of the Bayesian network, and obtaining the optimal Bayesian network structure includes the following steps:

[0033] S12451, Bayesian information criterion scoring for the complete structure of the Bayesian network;

[0034] S12452. Based on the greedy search strategy, the complete structure of the Bayesian network is repeatedly subjected to edge addition, edge reduction and edge rotation operations, and the direction of edge rotation is determined according to the mutual information corresponding to the mutual information chain, so as to optimize the complete structure of the Bayesian network;

[0035] S12453. Determine the scoring result of the Bayesian information criterion. When the scoring result reaches the optimal value, no more searching is performed, and the optimal Bayesian network structure is obtained.

[0036] Preferably, collecting bladder cancer medical data, performing correlation analysis between bladder cancer and biomarkers, and establishing a correlation database comprises the following steps:

[0037] S21. Collect the diagnosis and treatment data of patients with bladder cancer and non-bladder cancer who have been clinically diagnosed with bladder cancer and the detection data of various biological markers;

[0038] S22. Analysis of the correlation between bladder cancer and biomarkers based on the π value rule;

[0039] S23. Find the biomarker combination with the highest correlation with the diagnosis of bladder cancer;

[0040] S24. The bladder cancer diagnosis results are stored together with the biomarker combination with the highest correlation therewith to form a correlation database.

[0041] Preferably, the correlation analysis between bladder cancer and biomarkers based on the π value rule comprises the following steps:

[0042] S221. Select biomarker indicators for bladder cancer correlation with biomarkers;

[0043] S222. For each biomarker indicator, calculating the correlation status value between bladder cancer patients and non-bladder cancer patients;

[0044] S223, screening out biomarkers corresponding to association status values ​​within a preset threshold range;

[0045] S224. Establishing a logistic regression model between the screened biomarkers and the diagnosis results of bladder cancer;

[0046] S225. Validate the established logistic regression model, observe the effect of the screened biomarkers for bladder cancer diagnosis, and adjust the threshold range according to the effect;

[0047] S226. Repeat steps S223-S225 until the combination of biomarkers with the highest correlation with the bladder cancer diagnosis result is found.

[0048] Preferably, for each biomarker index, calculating the correlation status value between bladder cancer patients and non-bladder cancer patients comprises the following steps:

[0049] S2221, normalizing the data of each biomarker indicator;

[0050] S2222. Draw a unit circle with a radius of 1, and divide the unit circle into sectors according to the number of biomarkers, wherein the sectors represent biomarker indicators;

[0051] S2223, on the edge of each sector area, dot the corresponding position according to the normalized index value, and connect all the points in sequence to form a closed figure;

[0052] S2224, calculating the association status values ​​of bladder cancer patients and non-bladder cancer patients according to the association status value calculation formula;

[0053] The calculation formula of the association status value is:

[0054] W = (s1 + s2) / π;

[0055] Where W represents the association status value between bladder cancer patients and non-bladder cancer patients;

[0056] s1 represents the sum of the sectors in the closed figure;

[0057] s2 represents the sum of the areas of triangles in the closed graph;

[0058] π stands for pi.

[0059] Preferably, according to the determined biomarkers, collecting the patient's test sample and performing biomarker extraction comprises the following steps:

[0060] S41. Determine the patient test samples that need to be collected based on the determined biomarkers;

[0061] S42, extracting biomarkers from the patient according to the patient test samples that need to be collected;

[0062] S43. Detect the extracted biomarkers using a detection method.

[0063] The beneficial effects of the present invention are:

[0064] 1. The present invention can determine possible biomarkers early by collecting patients' medical records and diagnosing bladder cancer based on diagnostic models, thereby reducing invalid or redundant tests and making the diagnostic process more efficient. Through association analysis technology, the relationship between bladder cancer and biomarkers is analyzed in detail, and an association database is established, which can improve the accuracy of biomarker detection and the accuracy of bladder cancer diagnosis. Based on the bladder cancer diagnosis results and the association database, the biomarkers associated with the patient's bladder cancer can be determined, which will help to formulate more personalized and targeted treatment plans. Finally, according to the determined biomarkers, only relevant samples need to be collected for testing, thereby reducing the number of samples that need to be tested and saving time and resources.

[0065] 2. The bladder cancer diagnosis model constructed by the present invention through data preprocessing steps, selection of Bayesian network nodes and application of structure optimization algorithm can provide highly accurate diagnosis results. By collecting and analyzing multiple aspects of patient information, a comprehensive patient health status assessment can be provided, thereby more accurately identifying the possibility of bladder cancer. In the process of constructing and optimizing the Bayesian network, the most influential features can be found, thereby more effectively utilizing resources and reducing unnecessary testing and sampling. The diagnosis model can be used to automatically diagnose bladder cancer, reducing the burden on doctors and thereby improving the efficiency and effectiveness of medical services.

[0066] 3. The present invention can obtain large-scale clinical information by collecting the diagnosis and treatment data and biomarker detection data of historical clinically diagnosed bladder cancer patients and non-bladder cancer patients, and adopts the π value rule to evaluate the correlation between bladder cancer and biomarkers. It takes into account the normalization processing of biomarker indicators, draws unit circle graphics, calculates the correlation state value, and screens out the biomarkers with the highest correlation with the bladder cancer diagnosis results, which helps to identify those biomarkers with the greatest potential for bladder cancer. Establishing a logistic regression model is an effective method that can be used to associate the screened biomarkers with the bladder cancer diagnosis results, which helps to improve the accuracy of early diagnosis of bladder cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0068] Figure 1 The figure is a flow chart of a method for extracting bladder cancer biomarkers based on a diagnostic model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0069] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0070] According to an embodiment of the present invention, a method for extracting bladder cancer biomarkers based on a diagnostic model is provided.

[0071] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the method for extracting bladder cancer biomarkers based on a diagnostic model according to an embodiment of the present invention, the method for extracting bladder cancer biomarkers based on a diagnostic model comprises the following steps:

[0072] S1. Collect the patient's medical history information and diagnose bladder cancer for the patient based on the bladder cancer diagnosis model.

[0073] As a preferred embodiment, collecting the patient's medical history information and diagnosing bladder cancer for the patient based on the bladder cancer diagnosis model includes the following steps:

[0074] S11. Collect the patient's medical history information, which includes the patient's personal medical history, family medical history, symptoms, and imaging examination results.

[0075] For example, personal medical history includes any past diseases, illnesses, surgeries, medication use, etc., family medical history includes whether the patient's relatives have a history of bladder cancer or other related diseases, symptoms include common symptoms of bladder cancer, such as hematuria, frequent urination, urgency, pain, etc., and imaging examination results include all relevant medical imaging results, such as ultrasound, CT, MRI, etc.

[0076] S12. Construct a bladder cancer diagnosis model based on structure optimization algorithm.

[0077] Specifically, building a bladder cancer diagnosis model based on a structure optimization algorithm includes the following steps:

[0078] S121. Extract diagnostic data of bladder patients and non-bladder patients from the hospital medical record system to form a data set.

[0079] For example, diagnostic data for bladder patients and non-bladder patients include personal medical history, family medical history, symptoms, and imaging test results.

[0080] S122, preprocessing the formed data set, and using the preprocessed data set as a training data set, the preprocessing includes data cleaning, abnormal data deletion and missing value supplementation.

[0081] It should be noted that data cleaning includes identifying and processing duplicate data, checking and correcting the format of data, and verifying the accuracy of data.

[0082] Abnormal data refers to data that deviates significantly from other observed values. Abnormal data may be caused by input errors, equipment failures, or other unknown factors, which may have an adverse effect on the analysis results and therefore usually need to be deleted.

[0083] Missing value supplementation means that in the process of collecting and recording data, some data may be lost or not recorded. Missing values ​​can be supplemented through machine learning algorithms.

[0084] S123. Determine variable information related to bladder diagnosis based on medical literature, select the variable information from the training data set and use it as a Bayesian network node.

[0085] It should be noted that the variable information includes the patient's age, gender, family medical history, smoking history, symptoms, laboratory test results, imaging test results, etc. After determining the relevant variables, these variables are selected from the training data set, and each selected variable is used as a node in the Bayesian network; the Bayesian network is a graph model in which nodes represent variables and edges represent the dependencies between variables.

[0086] S124. Construct a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structural optimization algorithm.

[0087] Specifically, constructing a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structural optimization algorithm includes the following steps:

[0088] S1241. For each pair of Bayesian network nodes, calculate the mutual information between them.

[0089] It should be noted that mutual information is a measure of the degree of dependence between two random variables, that is, the amount of information one variable contains about another variable, which specifically includes the following steps:

[0090] Step 1: Create a joint distribution table, list all possible combinations of values ​​of the two nodes and, and record the number of times they appear at the same time.

[0091] Step 2: Calculate the marginal distribution of each node, that is, the probability distribution of two nodes appearing alone.

[0092] Step 3: Use the joint distribution and marginal distribution to calculate the mutual information, which is the ratio of the joint distribution to the marginal distribution, then take the logarithm of the ratio and multiply it by the joint distribution.

[0093] Step 4: Sum the results of step 3 to obtain the mutual information between the two nodes.

[0094] S1242. Arrange the calculated mutual information in descending order, and add edges in sequence to construct an initial structure of the Bayesian network.

[0095] Specifically, the calculated mutual information is arranged in descending order, and edges are added in sequence. Constructing the initial structure of the Bayesian network includes the following steps:

[0096] S12421. Arrange the calculated mutual information in descending order to obtain a mutual information chain.

[0097] Specifically, the calculated results are arranged in order from large to small, and according to this order, random variable pairs with large mutual information are added one by one, so that the mutual information chain of the entire random variable set can be gradually constructed.

[0098] S12422. According to the obtained mutual information chain, the node pairs are taken out in sequence, and it is determined whether there is a path between the node pairs.

[0099] Specifically, starting from the first node pair of the mutual information chain, the two nodes are recorded, and it is determined whether the two recorded nodes will appear again in the subsequent node pairs. If so, there is a path between the two nodes through this common node. If the two nodes do not appear again in the subsequent node pairs, it is considered that there is no direct path between them.

[0100] S12423. If there is a path between the node pairs, an undirected edge is added between the node pairs, and the node pairs to which the undirected edge is added are deleted from the mutual information chain.

[0101] It should be noted that if there is a path between two nodes, it means that the two nodes may have a direct or indirect dependency relationship. In this case, an undirected edge can be added between the two nodes to represent this relationship. The undirected edge indicates a direct connection or interaction between the two nodes, but this connection or interaction has no clear directionality.

[0102] S12424. Repeat steps S12422-S12423 until the mutual information chain is empty.

[0103] S1243. Perform a conditional independence test on the nodes without edges added in the initial structure of the Bayesian network, and determine whether there are implicit edges between the nodes without edges added. If so, add the implicit edges to the initial structure of the Bayesian network to obtain a secondary structure of the Bayesian network.

[0104] It should be noted that the conditional independence test is a method to determine whether two random variables are independent given a third random variable. It is often used in the structural learning of Bayesian networks to determine whether an edge should be added between two nodes in the network.

[0105] S1244. Perform isolated node detection on the secondary structure of the Bayesian network, calculate the mutual information between the isolated node and other nodes, connect the node with the largest mutual information with the isolated node with an undirected edge, and obtain the complete structure of the Bayesian network.

[0106] It should be noted that by traversing the secondary structure of the Bayesian network and detecting whether there are isolated nodes, that is, nodes with a degree of 0, the complete structure of the Bayesian network can be obtained by calculating the mutual information between nodes and connecting the isolated nodes in a targeted manner.

[0107] S1245. Perform structural search on the complete structure of the Bayesian network through a structural optimization algorithm, optimize the edges of the complete structure of the Bayesian network, obtain the optimal Bayesian network structure, and use it as a bladder cancer diagnosis model.

[0108] Specifically, the complete structure of the Bayesian network is searched by a structure optimization algorithm, and the edges of the complete structure of the Bayesian network are optimized to obtain the optimal Bayesian network structure, which includes the following steps:

[0109] S12451, Bayesian network complete structure for Bayesian information criterion scoring.

[0110] It should be noted that the Bayesian Information Criterion is a scoring method for evaluating the results of Bayesian network structure learning. It provides an objective function for structure learning. By optimizing the score, the optimal Bayesian network structure can be learned.

[0111] S12452. Based on the greedy search strategy, the complete structure of the Bayesian network is repeatedly subjected to edge addition, edge reduction and edge rotation operations, and the direction of edge rotation is determined according to the mutual information corresponding to the mutual information chain, so as to optimize the complete structure of the Bayesian network.

[0112] It should be noted that the greedy search strategy is a common heuristic algorithm for optimization and search problems. Its basic idea is: in each step of selection, choose the option that seems to be the best at the moment, without considering possible choices in the future, and only make the local optimal (not the global optimal) choice, and repeat this process until a feasible solution to the problem is obtained.

[0113] S12453. Determine the scoring result of the Bayesian information criterion. When the scoring result reaches the optimal value, no more searching is performed, and the optimal Bayesian network structure is obtained.

[0114] S13. Use the patient's medical history information as input and input it into the bladder cancer diagnosis model, and output the patient's bladder cancer diagnosis result.

[0115] Specifically, the patient's medical records are collected, including age, gender, smoking history, symptoms, laboratory test indicators and other information related to bladder cancer diagnosis. The bladder cancer diagnosis model outputs the patient's bladder cancer diagnosis results based on the input medical record information and the knowledge learned during the training process.

[0116] Specifically, through data preprocessing steps, selection of Bayesian network nodes and application of structural optimization algorithms, the constructed bladder cancer diagnosis model can provide highly accurate diagnostic results. By collecting and analyzing multiple aspects of patient information, it can provide a comprehensive assessment of the patient's health status, thereby more accurately identifying the possibility of bladder cancer. In the process of constructing and optimizing the Bayesian network, the most influential features can be found, thereby more effectively utilizing resources and reducing unnecessary testing and sampling. The diagnostic model can be used to automatically diagnose bladder cancer, reducing the burden on doctors and thereby improving the efficiency and effectiveness of medical services.

[0117] S2. Collect medical data on bladder cancer, conduct correlation analysis between bladder cancer and biomarkers, and establish a correlation database.

[0118] As a preferred embodiment, collecting bladder cancer medical data, and performing correlation analysis between bladder cancer and biomarkers, and establishing a correlation database includes the following steps:

[0119] S21. Collect the diagnosis and treatment data of patients with clinically confirmed bladder cancer and non-bladder cancer and the detection data of various biological markers.

[0120] Specifically, the diagnosis and treatment data include the reasons for bladder cancer diagnosis and treatment, the results of bladder cancer diagnosis, etc.

[0121] S22. Analyze the correlation between bladder cancer and biomarkers based on the π value rule.

[0122] Specifically, the correlation analysis between bladder cancer and biomarkers based on the π value rule includes the following steps:

[0123] S221. Select biomarker indicators that correlate bladder cancer with biomarkers.

[0124] Specifically, we consulted relevant literature and selected biomarkers that have been reported to be potentially associated with bladder cancer as candidate indicators, such as serum tumor markers, urine tumor markers, etc. We also referred to the occurrence and development mechanism of bladder cancer and selected potential biomarkers based on biological processes related to the mechanism, such as genes or proteins related to cell apoptosis and tumor metastasis. We consulted experts and selected biomarker indicators related to the diagnosis and prognosis of bladder cancer based on clinical experience.

[0125] S222. For each biomarker indicator, calculate the correlation status value between bladder cancer patients and non-bladder cancer patients.

[0126] Specifically, for each biomarker indicator, calculating the association status value between bladder cancer patients and non-bladder cancer patients includes the following steps:

[0127] S2221. Normalize the data of each biomarker indicator to convert it into a dimensionless value between 0 and 1.

[0128] S2222. Draw a unit circle with a radius of 1, and divide the unit circle into sectors by installing the number of biomarkers, and the sectors represent biomarker indicators.

[0129] S2223. On the edge of each sector area, dot the corresponding position according to the normalized index value, and connect all the points in sequence to form a closed figure.

[0130] S2224. Calculate the association status values ​​of bladder cancer patients and non-bladder cancer patients according to the association status value calculation formula.

[0131] The calculation formula of the association status value is:

[0132] W=(s1+s2) / π.

[0133] Where W represents the association state value between bladder cancer patients and non-bladder cancer patients, s1 represents the sum of the fan-shaped areas in the closed figure, s2 represents the sum of the triangle areas in the closed figure, and π represents pi.

[0134] It should be noted that the triangular area in the closed figure refers to: a closed figure formed by marking and connecting the index values ​​of each biomarker on the unit circle. In addition to being composed of fan-shaped areas, this figure also contains some triangular areas. These triangular areas mainly appear between two adjacent index values ​​and between the first and last two index values.

[0135] S223. Screen out biomarkers corresponding to association status values ​​within a preset threshold range.

[0136] S224. Establish a logistic regression model between the screened biomarkers and bladder cancer diagnosis results.

[0137] S225. Verify the established logistic regression model, observe the effect of the screened biomarkers for bladder cancer diagnosis, and adjust the threshold range according to the effect.

[0138] S226. Repeat steps S223-S225 until the combination of biomarkers with the highest correlation with the bladder cancer diagnosis result is found.

[0139] S23. Find the biomarker combination that is most correlated with the diagnosis of bladder cancer.

[0140] S24. The bladder cancer diagnosis results are stored together with the biomarker combination with the highest correlation therewith to form a correlation database.

[0141] Specifically, by collecting the diagnosis and treatment data and biomarker detection data of historical clinically diagnosed bladder cancer patients and non-bladder cancer patients, large-scale clinical information can be obtained, and the π value rule is used to evaluate the correlation between bladder cancer and biomarkers. The normalization of biomarker indicators is considered, unit circle graphs are drawn, and the association state values ​​are calculated. The biomarkers with the highest correlation with the bladder cancer diagnosis results are screened out, which helps to identify those biomarkers with the greatest potential for bladder cancer. Establishing a logistic regression model is an effective method that can be used to associate the screened biomarkers with the bladder cancer diagnosis results, which helps to improve the accuracy of early diagnosis of bladder cancer.

[0142] S3. Determine biomarkers associated with the patient's bladder cancer based on the patient's bladder cancer diagnosis results and related databases.

[0143] It should be noted that the database is queried to obtain all stored bladder cancer-related biomarkers, and the corresponding bladder cancer diagnosis model in the database is found according to the patient's bladder cancer diagnosis results. Among the biomarkers contained in the diagnosis model, the top N biomarkers that contribute the greatest diagnostic evidence value to the patient are identified.

[0144] S4. Based on the determined biomarkers, collect the patient's test samples and perform biomarker extraction.

[0145] As a preferred embodiment, according to the determined biomarkers, collecting the patient's test sample and extracting the biomarkers includes the following steps:

[0146] S41. Determine the patient test samples that need to be collected based on the determined biomarkers.

[0147] S42. Extract biomarkers from patients based on the patient test samples that need to be collected.

[0148] S43. Detect the extracted biomarkers using a detection method.

[0149] For example, if you need to detect proteins or peptides in serum or cell culture supernatant, you can choose enzyme-linked immunosorbent assay, which is easy to operate, highly sensitive, and can be used for quantitative detection. If you need to detect the expression level of mRNA in cells, you can choose polymerase chain reaction, which can amplify and detect very low copy number DNA with high sensitivity. If you need to detect protein expression and relative molecular mass, you can choose Western blotting, which can perform qualitative and semi-quantitative detection.

[0150] In summary, with the help of the above technical solutions of the present invention, the present invention can determine possible biomarkers in advance by collecting patients' medical records and diagnosing bladder cancer based on diagnostic models, thereby reducing invalid or redundant tests and making the diagnostic process more efficient. Through association analysis technology, the relationship between bladder cancer and biomarkers is analyzed in detail, and an associated database is established, which can improve the accuracy of biomarker detection and the accuracy of bladder cancer diagnosis. Based on the bladder cancer diagnosis results and the associated database, biomarkers associated with the patient's bladder cancer can be determined, which will help to formulate more personalized and targeted treatment plans. Finally, according to the determined biomarkers, only relevant samples need to be collected for testing, thereby reducing the number of samples that need to be tested and saving time and resources. The present invention constructs a bladder cancer diagnostic model through data preprocessing steps, selection of Bayesian network nodes and application of structural optimization algorithms, which can provide highly accurate diagnostic results. By collecting and analyzing multiple aspects of patient information, a comprehensive The health status of patients can be assessed, so as to more accurately identify the possibility of bladder cancer. In the process of constructing and optimizing the Bayesian network, the most influential features can be found, so as to more effectively utilize resources and reduce unnecessary testing and sampling. The diagnosis model can be used to automatically diagnose bladder cancer, reduce the burden on doctors, and thus improve the efficiency and effectiveness of medical services. The present invention can obtain large-scale clinical information by collecting the diagnosis and treatment data and biomarker detection data of historical clinically diagnosed bladder cancer patients and non-bladder cancer patients, and adopts the π value rule to evaluate the correlation between bladder cancer and biomarkers, considers the normalization processing of biomarker indicators, draws unit circle graphics, calculates the association state value, and screens out the biomarkers with the highest correlation with the bladder cancer diagnosis results, which helps to identify those biomarkers with the greatest potential for bladder cancer. Establishing a logistic regression model is an effective method that can be used to associate the screened biomarkers with the bladder cancer diagnosis results, which helps to improve the accuracy of early diagnosis of bladder cancer.

[0151] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0152] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for extracting bladder cancer biomarkers based on a diagnostic model, characterized in that: The method for extracting bladder cancer biomarkers based on the diagnostic model includes the following steps: S1. Collect the patient's medical history information and analyze the patient's bladder cancer data based on the bladder cancer diagnosis model; S2. Collect medical data on bladder cancer, conduct correlation analysis between bladder cancer and biomarkers, and establish a correlation database; S3. determining a biomarker associated with bladder cancer of the patient based on the patient's bladder cancer diagnosis result and a related database; S4. Based on the determined biomarkers, collect the patient's test samples and perform biomarker extraction; The collecting of the patient's medical history information and performing bladder cancer data analysis on the patient based on the bladder cancer diagnosis model comprises the following steps: S11. Collect the patient's medical history information, including the patient's personal medical history, family medical history, symptoms, and imaging examination results; S12. Construction of bladder cancer diagnosis model based on structure optimization algorithm; S13, taking the patient's medical record information as input and inputting it into the bladder cancer diagnosis model, and outputting the patient's bladder cancer result; The construction of a bladder cancer diagnosis model based on a structure optimization algorithm comprises the following steps: S121, extracting diagnostic data of bladder patients and non-bladder patients from a hospital medical record system to form a data set; S122, preprocessing the formed data set, and using the preprocessed data set as a training data set, wherein the preprocessing includes data cleaning, abnormal data deletion, and missing value supplementation; S123, determining variable information related to bladder diagnosis according to medical literature, selecting the variable information from the training data set and using it as a Bayesian network node; S124. Constructing a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structural optimization algorithm; The method of constructing a bladder cancer diagnosis model based on the determined Bayesian network nodes and using a structural optimization algorithm comprises the following steps: S1241. For each pair of Bayesian network nodes, calculate the mutual information between them; S1242, arranging the calculated mutual information in descending order, and adding edges in sequence to construct an initial structure of the Bayesian network; S1243, performing a conditional independence test on the nodes without edges added in the initial structure of the Bayesian network, and determining whether there are implicit edges between the nodes without edges added, and if so, adding the implicit edges to the initial structure of the Bayesian network to obtain a secondary structure of the Bayesian network; S1244, performing isolated node detection on the secondary structure of the Bayesian network, and calculating the mutual information between the isolated node and other nodes, connecting the node with the largest mutual information with the isolated node with an undirected edge, and obtaining the complete structure of the Bayesian network; S1245, performing a structural search on the complete structure of the Bayesian network by using a structural optimization algorithm, optimizing the edges of the complete structure of the Bayesian network, obtaining an optimal Bayesian network structure, and using it as a bladder cancer diagnosis model; The calculated mutual information is arranged in descending order, and edges are added in sequence to construct the initial structure of the Bayesian network, including the following steps: S12421. Arrange the calculated mutual information in descending order to obtain a mutual information chain; S12422. According to the obtained mutual information chain, node pairs are taken out in sequence, and it is determined whether there is a path between the node pairs; S12423. If there is a path between the node pairs, an undirected edge is added between the node pairs, and the node pairs to which the undirected edge is added are deleted from the mutual information chain; S12424. Repeat steps S12422-S12423 until the mutual information chain is empty; The method of performing a structural search on the complete structure of the Bayesian network by using a structural optimization algorithm, optimizing the edges of the complete structure of the Bayesian network, and obtaining the optimal Bayesian network structure includes the following steps: S12451, Bayesian information criterion scoring for the complete structure of the Bayesian network; S12452. Based on the greedy search strategy, the complete structure of the Bayesian network is repeatedly subjected to edge addition, edge reduction and edge rotation operations, and the direction of edge rotation is determined according to the mutual information corresponding to the mutual information chain, so as to optimize the complete structure of the Bayesian network; S12453. Determine the scoring result of the Bayesian information criterion. When the scoring result reaches the optimal value, no more searching is performed, and the optimal Bayesian network structure is obtained.

2. The method for extracting bladder cancer biomarkers based on a diagnostic model according to claim 1, characterized in that: The collecting of bladder cancer medical data, performing correlation analysis between bladder cancer and biomarkers, and establishing a correlation database comprises the following steps: S21. Collect the diagnosis and treatment data of patients with bladder cancer and non-bladder cancer who have been clinically diagnosed with bladder cancer and the detection data of various biological markers; S22. Analysis of the correlation between bladder cancer and biomarkers based on the π value rule; S23. Find the biomarker combination with the highest correlation with the diagnosis of bladder cancer; S24. The bladder cancer diagnosis results are stored together with the biomarker combination with the highest correlation therewith to form a correlation database.

3. The method for extracting bladder cancer biomarkers based on a diagnostic model according to claim 2, characterized in that: The correlation analysis between bladder cancer and biomarkers based on the π value rule includes the following steps: S221. Select biomarker indicators for bladder cancer correlation with biomarkers; S222. For each biomarker indicator, calculating the correlation status value between bladder cancer patients and non-bladder cancer patients; S223, screening out biomarkers corresponding to association status values ​​within a preset threshold range; S224. Establishing a logistic regression model between the screened biomarkers and the diagnosis results of bladder cancer; S225. Validate the established logistic regression model, observe the effect of the screened biomarkers for bladder cancer diagnosis, and adjust the threshold range according to the effect; S226. Repeat steps S223-S225 until the combination of biomarkers with the highest correlation with the bladder cancer diagnosis result is found.

4. The method for extracting bladder cancer biomarkers based on a diagnostic model according to claim 3, characterized in that: Calculating the correlation status value between bladder cancer patients and non-bladder cancer patients for each biomarker index comprises the following steps: S2221, normalizing the data of each biomarker indicator; S2222. Draw a unit circle with a radius of 1, and divide the unit circle into sectors according to the number of biomarkers, wherein the sectors represent biomarker indicators; S2223, on the edge of each sector area, dot the corresponding position according to the normalized index value, and connect all the points in sequence to form a closed figure; S2224, calculating the association status values ​​of bladder cancer patients and non-bladder cancer patients according to the association status value calculation formula; The calculation formula of the association status value is: W = (s1 + s2) / π; Where W represents the association status value between bladder cancer patients and non-bladder cancer patients; s1 represents the sum of the sectors in the closed figure; s2 represents the sum of the areas of triangles in the closed graph; π stands for pi.

5. The method for extracting bladder cancer biomarkers based on a diagnostic model according to claim 1, characterized in that: The method of collecting a patient's test sample and extracting the biomarker according to the determined biomarker comprises the following steps: S41. Determine the patient test samples that need to be collected based on the determined biomarkers; S42, extracting biomarkers from the patient according to the patient test samples that need to be collected; S43. Detect the extracted biomarkers using a detection method.

Citation Information

Patent Citations

  • Application of cfDNA in non-invasive diagnosis of early breast cancer

    CN111910004A

  • Deep learning-based diagnosis and referral of diseases and disorders using natural language processing

    CN113015977A

  • Clinical patient monitoring method and system and computer storage medium

    CN114300073A

  • Bladder cancer biomarker extraction method and bladder cancer diagnosis model

    CN115951067A

  • Bladder cancer methylation site marker and application thereof

    CN116479132A