Serum tumor marker detection array optimization method and system

Abstract

The invention provides a serum tumor marker detection array optimization method and system.The method is applied to a control module of an array device.The method comprises the steps that when a multi-needle detection platform conducts serum tumor marker detection on a preset sample, preliminary expression information sent by the multi-needle detection platform is obtained; acquiring patient medical history information, iconography information and pathology information corresponding to the preset sample; inputting the preliminary expression information, the patient medical history information, the iconography information and the pathology information into a pre-constructed Bayesian analysis model, and outputting a prediction weight, a prediction disease progress direction and a correlation marker corresponding to each tumor marker to generate target information; wherein the target spot information comprises a plurality of target markers; and controlling the multi-needle detection platform to perform serum tumor marker detection on the preset sample again according to the target spot information to complete array optimization of serum tumor marker detection. Therefore, an efficient and innovative solution is provided for early diagnosis and monitoring.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, which is applied in the medical field, and in particular to a method and system for optimizing a serum tumor marker detection array. Background Art

[0002] Since the 1960s, when alpha-fetoprotein (AFP) and carcinoembryonic antigen (CEA) were discovered and used in clinical testing, the concept of tumor markers has been widely accepted by the medical community and has been widely valued and studied by medical scientists around the world. Currently, there are hundreds of known tumor markers.

[0003] It is difficult to test all the above serum tumor markers at one time in most grassroots hospitals. The current test kits usually use ELISA method, which mainly includes alpha-fetoprotein, carcinoembryonic antigen, neuron-specific enolase, squamous cell carcinoma antigen, cytokeratin 19 fragment, gastrin-releasing peptide precursor, carbohydrate antigen 125, carbohydrate antigen 15-3 and carbohydrate antigen 19-9. The current test kits can generally only select 3-5 markers at the same time, and the follow-up times and compliance of grassroots patients are limited, making it difficult to take samples multiple times.

[0004] Therefore, it is urgent to design a technical solution to solve at least one of the above technical problems. Summary of the invention

[0005] The present application provides a serum tumor marker detection array optimization method and system, which aims to solve the problem that it is difficult to carry out the one-time detection of all the above serum tumor markers in the vast grassroots hospitals. The current detection kits usually adopt the ELISA method, which mainly includes alpha-fetoprotein, carcinoembryonic antigen, neuron-specific enolase, squamous cell carcinoma antigen, cytokeratin 19 fragment, gastrin-releasing peptide precursor, carbohydrate antigen 125, carbohydrate antigen 15-3 and carbohydrate antigen 19-9. The current detection kits can generally only select 3-5 markers at the same time, and the follow-up number and compliance of grassroots patients are limited, making it difficult to take samples multiple times.

[0006] In a first aspect, the present application provides a method for optimizing a serum tumor marker detection array, which is applied to a control module of an array device, wherein the array device further comprises a multi-needle detection platform and a microfluidic chip, wherein the multi-needle detection platform is used to simultaneously detect multiple biomarkers on the microfluidic chip, wherein a preset sample is placed in the microfluidic chip; the method comprises:

[0007] When the multi-needle detection platform performs serum tumor marker detection on the preset sample, obtaining preliminary expression information sent by the multi-needle detection platform; wherein the preliminary expression information includes multiple tumor markers and preliminary expression levels corresponding to each of the tumor markers, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen;

[0008] Obtaining patient medical history information, imaging information, and pathology information corresponding to the preset sample;

[0009] Inputting the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, the Bayesian analysis model outputs the prediction weight corresponding to each tumor marker, the predicted disease progression direction and the association marker;

[0010] Generate target information according to the prediction weight, the predicted disease progression direction and the associated markers; wherein the target information includes a plurality of target markers;

[0011] The multi-needle detection platform is controlled to re-detect the serum tumor markers on the preset sample according to the target point information, so as to complete the array optimization of the serum tumor marker detection.

[0012] In some embodiments, the Bayesian analysis model includes multiple nodes, a directed acyclic graph and a conditional probability table; before the preliminary expression information, patient medical history information, imaging information and pathological information are input into the pre-constructed Bayesian analysis model, the method also includes: using the preliminary expression information, patient medical history information, imaging information and pathological information as variable information of the Bayesian analysis model to be constructed; constructing a node corresponding to each variable information in the Bayesian analysis model to be constructed; obtaining dependency information of each node with the remaining nodes; generating a directed acyclic graph in the Bayesian analysis model according to the dependency information corresponding to each node; obtaining a historical data set corresponding to the serum tumor marker detection; generating the conditional probability table according to a preset estimation algorithm and the historical data set to complete the construction of the Bayesian analysis model.

[0013] In some embodiments, before inputting the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, the method further includes: one-hot encoding the patient medical history information to obtain first feature information; performing principal component analysis on the pathological information to obtain second feature information; performing image feature extraction on the imaging information to obtain third feature information; normalizing the first feature information, the second feature information and the third feature information; and inputting the first feature information, the second feature information and the third feature information into the Bayesian analysis model to obtain the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker.

[0014] In some embodiments, the preliminary expression information, patient medical history information, imaging information and pathological information are input into a pre-constructed Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker, including: respectively inputting the preliminary expression information, patient medical history information, imaging information and pathological information into corresponding nodes in the Bayesian analysis model, and the Bayesian analysis model calculates the posterior probability distribution information corresponding to each tumor marker in the preliminary expression information based on a preset Bayesian inference algorithm; and obtaining the prediction weight, predicted disease progression direction and correlation marker according to the posterior probability distribution information; wherein the Bayesian inference algorithm includes any one of variational inference and Markov chain Monte Carlo method.

[0015] Exemplarily, the obtaining of the prediction weight, the predicted disease progression direction and the associated marker according to the posterior probability distribution information includes: obtaining the prediction weight according to the posterior probability distribution information; forming a weight vector according to a plurality of the prediction weights; obtaining the concentration information corresponding to each tumor marker according to the preliminary expression level corresponding to each tumor marker; obtaining a disease progression threshold according to the weight vector and the plurality of the concentration information; obtaining the predicted disease progression direction by comparing the disease progression threshold with a preset threshold; wherein the predicted disease progression direction includes any one of worsening, maintenance, remission and stability; and obtaining the associated marker in a preset marker database according to the weight vector.

[0016] In some embodiments, generating target target information based on the prediction weights, predicted disease progression direction and associated markers includes: determining an abnormal weight among multiple prediction weights; obtaining a weight distribution corresponding to the remaining prediction weights except the abnormal weight; correcting the abnormal weight according to the weight distribution; generating target target information based on the corrected prediction weights, predicted disease progression direction and associated markers.

[0017] Exemplarily, the modifying the abnormal weight according to the weight distribution includes: obtaining the standard deviation, mean and median corresponding to the remaining predicted weights except the abnormal weight according to the weight distribution; modifying the abnormal weight according to the standard deviation, mean and median, and the expression of the abnormal weight includes:

[0018] W'=k1*σ+k2*μ+k3*(ε-W);

[0019] Wherein, W' is the abnormal weight after repair, W is the abnormal weight, σ is the standard deviation, μ is the mean, ε is the median, k1 is the first coefficient, k2 is the second coefficient, and k3 is the third coefficient.

[0020] In some embodiments, generating target point information according to the prediction weight, predicted disease progression direction and associated markers includes: confirming a target weight among a plurality of the prediction weights; determining the tumor marker and the associated marker corresponding to the target weight as suspicious markers; calculating the influencing factors of each predicted disease progression direction and the suspicious marker; determining a target marker among a plurality of the suspicious markers according to the influencing factors, so as to form the target point according to the target marker; wherein the expression of each of the influencing factors is:

[0021] I ij =k j *P(D i |M j )*C ij ;

[0022] Among them, I ij The predicted disease development direction D for the i-th i and the jth suspected marker M j The influencing factors, k j is the weight of the jth target marker, C ij is the correlation coefficient between the i-th suspected marker and the j-th target marker, P(D i |M j ) is used to describe the suspected marker M j The following disease development direction D i The corresponding conditional probability.

[0023] In a second aspect, an embodiment of the present application provides an array device, which includes a control module, a multi-needle detection platform and a microfluidic chip, wherein the multi-needle detection platform is used to simultaneously detect multiple biomarkers on the microfluidic chip, and a preset sample is placed in the microfluidic chip; wherein the control module includes a memory and a processor; the memory is used to store a computer program; and the processor is used to execute the computer program and implement the serum tumor marker detection array optimization method as described in the first aspect when executing the computer program.

[0024] In a third aspect, an embodiment of the present application provides a serum tumor marker detection array optimization system, which is applied to a control module of an array device provided in any embodiment of the present application; the system comprises:

[0025] A preliminary detection unit, used for obtaining preliminary expression information sent by the multi-needle detection platform when the multi-needle detection platform detects serum tumor markers on the preset sample; wherein the preliminary expression information includes a plurality of tumor markers and a preliminary expression level corresponding to each of the tumor markers, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen;

[0026] An information acquisition unit, used to acquire the patient's medical history information, imaging information and pathology information corresponding to the preset sample;

[0027] An information input unit, used to input the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, wherein the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker;

[0028] A target generating unit, used to generate target information according to the prediction weight, the predicted disease progression direction and the associated marker; wherein the target information includes a plurality of target markers;

[0029] The array optimization unit is used to control the multi-needle detection platform to re-detect the serum tumor markers on the preset samples according to the target information, so as to complete the array optimization of the serum tumor marker detection.

[0030] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the serum tumor marker detection array optimization method as described in the first aspect.

[0031] The present application provides a method and system for optimizing a serum tumor marker detection array. Both the method and the system are applied to a control module of an array device. The array device also includes a multi-needle detection platform and a microfluidic chip. The multi-needle detection platform is used to simultaneously detect multiple biomarkers on the microfluidic chip, and a preset sample is placed in the microfluidic chip. The method and the system use a multi-needle detection platform and a microfluidic chip in combination, and can simultaneously detect multiple biomarkers at a microscale. The preset sample is detected on the microfluidic chip by the multi-needle detection platform to obtain the preliminary expression levels of multiple tumor markers including alpha-fetoprotein and carcinoembryonic antigen. In addition to marker detection, the patient's medical history information, imaging information and pathological information are also combined to form a more comprehensive diagnosis basis. The above information is input into a pre-constructed Bayesian analysis model. The model outputs the predicted weight of each tumor marker, the predicted disease progression direction and the correlation marker according to the input data. Based on the results of the Bayesian analysis, a set of optimized target marker lists is generated. The multi-needle detection platform re-detects the sample according to the target target information to optimize the detection strategy. Furthermore, the provided method and system are of great significance for improving the capabilities of primary hospitals in tumor detection, especially under resource-constrained conditions, and provide an efficient and innovative solution for early diagnosis and monitoring.

[0032] At the same time, the provided method also has the following beneficial effects:

[0033] 1. Improve detection efficiency: A single test can cover more tumor markers, reduce the number of tests, and improve patient compliance.

[0034] 2. Data-driven optimization: With the help of Bayesian analysis models, personalized marker optimization is achieved by integrating patient medical history and test results, which is more targeted and accurate.

[0035] 3. Resource conservation: By optimizing the marker selection and detection process, resources can be effectively utilized, which is particularly suitable for application in primary medical institutions.

[0036] 4. Improve diagnostic accuracy: Integrating multimodal information and multi-marker data can provide more accurate disease assessment and prediction compared to a single method.

[0037] 5. Reduce the burden on patients: Reduce the need for frequent sampling, making the patient's medical treatment process more convenient and comfortable.

[0038] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 A schematic block diagram of the structure of an array device shown in an embodiment of the present application;

[0041] Figure 2 This is a schematic flow chart of the steps of the serum tumor marker detection array optimization method shown in the embodiment of the present application;

[0042] Figure 3 A schematic diagram of the structure of the system shown in the embodiment of the present application;

[0043] Figure 4 This is a schematic block diagram of the structure of the serum tumor marker detection array optimization system shown in the embodiment of the present application.

[0044] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. DETAILED DESCRIPTION

[0045] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, systems, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0046] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0047] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0048] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0049] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0050] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0051] The technical solution of the embodiment of the present application is introduced below.

[0052] Please refer to Figure 1 , Figure 1 1 is a schematic block diagram of the structure of an array device 10 provided in an embodiment of the present application. Figure 1As shown, the array device 10 includes a multi-needle detection platform 11, a microfluidic chip 12 and a control module 13. The multi-needle detection platform 11 is used to simultaneously detect multiple biomarkers on the microfluidic chip 12, and a preset sample is placed in the microfluidic chip 12. The provided control module 13 is used to implement the precise positioning method of the catheter tip provided in any embodiment of the present application. The method is to obtain preliminary expression information sent by the multi-needle detection platform when the multi-needle detection platform detects serum tumor markers on preset samples; wherein the preliminary expression information includes multiple tumor markers and the preliminary expression level corresponding to each tumor marker, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen; obtain patient medical history information, imaging information and pathological information corresponding to the preset sample; input the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-constructed Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker; generate target point information according to the prediction weight, predicted disease progression direction and correlation marker; wherein the target target information includes multiple target markers; control the multi-needle detection platform to re-detect serum tumor markers on the preset sample according to the target target information, and complete the array optimization of serum tumor marker detection.

[0053] Specifically, the multi-needle detection platform 11 is composed of a plurality of independent needles 111 ( Figure 1 The multi-needle detection platform 11 is composed of a plurality of needles (only one is shown in the figure), which can contact multiple biomarkers in a preset sample on the microfluidic chip 12 at the same time. Each needle 111 is connected to a sensor for detecting a specific biomarker reaction signal. The multi-needle detection platform 11 performs sample detection through the microfluidic chip 12. It can contact multiple locations at the same time to achieve simultaneous detection of multiple markers, thereby improving detection efficiency. This requires precise positioning and control to ensure that the probe operates in the appropriate area on the microfluidic chip 12.

[0054] The microfluidic chip 12 includes a plurality of tiny channels and chambers, which are used to guide and accommodate sample flow. The surface of the microfluidic chip 12 is usually coated with specific antibodies or reagents for reacting with target biomarkers. After the preset sample is introduced into the microfluidic chip, it flows through the microchannels and reacts with the fixed markers. This reaction will produce a detectable signal (e.g., a fluorescent or electrochemical signal), which is picked up and analyzed by the multi-needle detection platform 11.

[0055] The control module 13 is composed of a microcontroller or microprocessor, an interface circuit, a signal processing unit, and a data communication module. For example, the Bayesian analysis model is also integrated in the control module 13. The control module 13 is responsible for coordinating the operation of the multi-needle detection platform 11 and the microfluidic chip 12. One of its key functions is to process the test data and combine the patient's medical history, imaging and pathology information, and use the Bayesian analysis model for comprehensive analysis. Based on the analysis results, the module generates target information and guides further biomarker detection.

[0056] Through the parallel operation of the multi-needle detection platform, multiple biomarkers can be detected simultaneously, which improves the detection speed and efficiency. This device can provide fast and accurate small molecule biomarker detection in clinical diagnosis, which is helpful for early disease screening and personalized treatment plan formulation.

[0057] In some embodiments, the control module 13 is a microcontroller unit (MCU), and any embodiment corresponding to the method provided in the present application can be implemented by the MCU.

[0058] In some embodiments, Figure 1 As shown, the control module 13 is connected to the multi-needle detection platform 11 in communication, and a control signal channel is set up between the control module 13 and the communication module built into the multi-needle detection platform 11 to control the probe operation life. At the same time, a data return path is provided to send the sensor signal corresponding to the multi-needle detection platform 11 back to the control module 13 for real-time analysis. The control module 13 sends instructions to the multi-needle detection platform 11 through a command parser to perform specific actions (such as probe movement, position calibration, etc.). At the same time, a shared clock or clock information embedded in the protocol can also be used for time synchronization to ensure the synchronization of data acquisition and control signals. The data flow is managed through a ring buffer or queue to ensure no omissions or delays, while supporting high-throughput data transmission.

[0059] In some embodiments, a plurality of micro needles 111 and a plurality of sensors are integrated on the multi-needle detection platform 11. Each needle 111 can be connected to a detection signal individually. These needles 111 are made by micro-fabrication technology and are suitable for high-density arrangement. The needles 111 are functionalized on the surface to carry specific binding antibodies or receptors to identify specific biomarkers.

[0060] Each needle 111 is also connected to a corresponding sensor for converting the biochemical reaction signal into a measurable electrical signal. This may include an electrochemical sensor (such as potential, current measurement) or an optical sensor (such as fluorescence, chemiluminescence detection). An amplifier and a filter circuit can also be built into the multi-needle detection platform 11 to increase the sensitivity of the signal and reduce noise.

[0061] Exemplarily, the multi-needle detection platform 11 further includes a moving mechanism, which includes a servo motor or a stepper motor, and can accurately control the movement and positioning of the needle 111 on the surface of the microfluidic chip 12 to align with its microfluidic channel.

[0062] The preset sample with tumor markers is first loaded onto the microfluidic chip 12. During the flow, the sample specifically binds to the functionalized needles 111 on the needle array to form a stable complex. The reaction on each needle 111 is converted into an electrical signal (such as a change in current) or an optical signal (such as a change in fluorescence intensity). The probe records the signal changes in real time through the sensor unit, and these signal changes reflect the presence and concentration of biomarkers in the sample. The multi-needle detection platform 11 generates preliminary expression information including multiple tumor markers such as alpha-fetoprotein and carcinoembryonic antigen and their corresponding preliminary expression levels by summarizing the detection results of multiple probes (or directly sending them to the control module 13 for processing).

[0063] Please refer to Figure 2 , Figure 2 The present application provides a schematic flow chart of the steps of a serum tumor marker detection array optimization method. The serum tumor marker detection array optimization method of the present application embodiment can be applied to the control module of the array device provided in any embodiment of the present application, and the control module can be an MCU. Figure 2 As shown, the serum tumor marker detection array optimization method of this embodiment includes steps S101 to S105, which are described in detail as follows:

[0064] Step S101, when the multi-needle detection platform performs serum tumor marker detection on a preset sample, preliminary expression information sent by the multi-needle detection platform is obtained; wherein the preliminary expression information includes multiple tumor markers and the preliminary expression level corresponding to each tumor marker, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen.

[0065] Specifically, during the test, the multi-needle detection platform detects multiple tumor markers in the preset sample through physical and chemical reactions. The preliminary expression information obtained includes the names of these markers and their corresponding preliminary expression levels. Special attention is paid to common tumor markers such as alpha-fetoprotein (AFP) and carcinoembryonic antigen (CEA). For example, AFP, CEA and other related markers are detected simultaneously from patient serum samples, and microfluidics technology and sensor detection are used to capture data and generate a preliminary report showing the concentration of each marker. Then, preliminary information of multiple tumor markers can be quickly obtained to facilitate subsequent analysis. Improve the speed of preliminary screening and reduce sample processing time through parallel detection.

[0066] Step S102, obtaining patient medical history information, imaging information and pathology information corresponding to a preset sample.

[0067] Specifically, by collecting complete patient information corresponding to the sample, including: Medical history information: past treatment and health status. Imaging information: such as CT or MRI results. Pathology information: tissue sample analysis. For example, extract relevant information from the hospital database through the data integration network in compliance with relevant regulations to ensure accurate interpretation in the context of the test results. Enhance the comprehensiveness of the analysis by providing detailed background information. Integrate multi-dimensional data to support personalized analysis.

[0068] Step S103, inputting the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight corresponding to each tumor marker, the predicted disease progression direction and the associated marker.

[0069] Specifically, the preliminary expression information and detailed information of the patient are input into the Bayesian analysis model. The model performs probabilistic reasoning and the output includes: the predicted weight of each marker. The predicted direction of disease progression. Associated markers to indicate potential biological correlations. As shown in the model output, the abnormally high expression of AFP and another marker in a patient together indicates a potential risk of progression to hepatocellular carcinoma. The probabilistic model is then used to integrate complex data to improve the accuracy of diagnostic predictions. Actionable results are provided to assist clinical decision-making.

[0070] In some embodiments, the Bayesian analysis model includes multiple nodes, a directed acyclic graph and a conditional probability table; before the preliminary expression information, patient medical history information, imaging information and pathological information are input into the pre-constructed Bayesian analysis model, the method also includes: using the preliminary expression information, patient medical history information, imaging information and pathological information as variable information of the Bayesian analysis model to be constructed; constructing a node corresponding to each variable information in the Bayesian analysis model to be constructed; obtaining dependency information of each node with the remaining nodes; generating a directed acyclic graph in the Bayesian analysis model according to the dependency information corresponding to each node; obtaining a historical data set corresponding to the serum tumor marker detection; generating the conditional probability table according to a preset estimation algorithm and the historical data set to complete the construction of the Bayesian analysis model.

[0071] Before inputting the data into the Bayesian analysis model, the preliminary expression information, patient history information, imaging information, and pathology information are used as variable information. This information will be used as nodes in the model to represent different factors that affect the results of tumor marker detection. Create corresponding nodes for each variable information. In the network of the Bayesian analysis model, nodes represent information about each variable or measurement (such as alpha-fetoprotein level, whether there is a chronic disease in the medical history, etc.). Analyze the dependency information from medical expertise and data analysis to determine the relationship between each node. These relationships describe how a node depends on the information of other nodes. Based on the dependency information of the nodes, a directed acyclic graph is constructed. This graph clearly describes the causal relationship between each variable and eliminates circular dependencies to ensure the feasibility of the analysis. Generate a conditional probability table for each node, defining the possibility of these node values ​​under different conditions. This step requires the use of historical data sets combined with preset estimation algorithms (such as maximum likelihood estimation or Bayesian estimation) to determine the probability value. Through the combination of variable nodes, DAG and CPT, a complete Bayesian analysis model is constructed for subsequent data input and inference.

[0072] The Bayesian analysis model can integrate multi-dimensional data, and thus better adapt to the complex relationships in biomarker detection. Especially in the medical field, the relationships between variables are complex, and such a customized model can better capture potential key factors. The use of Bayesian analysis can provide probabilistic inferences to help identify important markers and predict disease progression. In this way, the model can make reasonable predictions even in the case of incomplete information. At the same time, through modeling and reasoning, the model can identify the main target points, reduce unnecessary sample testing, accurately focus on key markers, and improve the detection accuracy and effectiveness of the entire process. And by comprehensively analyzing a variety of information about individual patients, the Bayesian analysis model can provide personalized and detailed predictions, which helps to achieve precision medicine. Medical decisions can be based on comprehensive probabilistic analysis, rather than relying solely on single-point information, to support more informed clinical diagnosis and treatment plans.

[0073] The above embodiment effectively applies the Bayesian network to the optimization of tumor marker detection through the construction of the Bayesian analysis model. Through the structured modeling and solution scheme, the detection and diagnosis process is made more scientific and practical, providing a powerful tool for clinical practice.

[0074] Exemplarily, after the Bayesian analysis model is constructed, its parameters can also be optimized by training the Bayesian analysis model using a training data set and adjusting the network parameters to better fit the data. The performance of the model is evaluated by a cross-validation method to ensure the robustness and generalization ability of the model. The model parameters are tuned according to the verification results, such as adjusting the conditional probability table or network structure to improve the model accuracy. At the same time, standard evaluation indicators (such as accuracy, recall, F1 score, etc.) are used to evaluate the performance of the model on the test data set. The Bayesian analysis model understands which variables have the greatest impact on the output results by measuring the sensitivity to changes in different input variables. In addition, a Bayesian analysis model suitable for serum tumor marker detection can be constructed and implemented, thereby providing strong model support for disease diagnosis and treatment.

[0075] In some embodiments, before inputting the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, the method further includes: one-hot encoding the patient medical history information to obtain first feature information; performing principal component analysis on the pathological information to obtain second feature information; performing image feature extraction on the imaging information to obtain third feature information; normalizing the first feature information, the second feature information and the third feature information; and inputting the first feature information, the second feature information and the third feature information into the Bayesian analysis model to obtain the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker.

[0076] By performing one-hot encoding on the patient's medical history information. This method transforms categorical data into binary vectors and is suitable for processing non-numerical categorical data, such as whether the patient has a history of a specific disease. For pathological information, principal component analysis (PCA) is performed to reduce the dimension of the data. This method helps to extract the most informative features and remove possible noise data to simplify the input data structure and speed up the calculation. Extracting meaningful image features from imaging information can be achieved through convolutional neural networks (CNN) or other image processing techniques. This process extracts features in the imaging data that are useful for detection, such as shape, texture, and edge information. By normalizing the first feature information (medical history), the second feature information (pathology), and the third feature information (image), the normalization can adjust the data to a common scale, eliminate the impact of the dimension, and improve the training and prediction effects of the model. The normalized feature information is input into the pre-built Bayesian analysis model. The Bayesian model uses these pre-processed feature information for reasoning and outputs the prediction weight of each tumor marker, the direction of disease progression, and the associated markers.

[0077] Then, through unique-hot encoding, principal component analysis and image feature extraction, different types of data (classification, continuous, image) can be converted into model input in the most suitable way, thereby improving the accuracy of the analysis. Principal component analysis reduces the data dimension and the complexity of model calculation, while retaining the most important information in the data, improving the performance and response time of the model. Data normalization ensures that each feature is analyzed on the same scale, preventing deviations caused by differences in data magnitude, thereby improving the stability and accuracy of the model. Image feature extraction and processing technology improves the efficiency of extracting key information from complex imaging data and increases the application function of the Bayesian model to imaging data. After comprehensive processing of multi-dimensional information, the Bayesian analysis model can provide more accurate prediction and diagnosis support, helping doctors make more accurate decisions on disease prediction and treatment plan selection.

[0078] Through these preprocessing and feature enhancement techniques, the above embodiments greatly improve the applicability and effectiveness of the Bayesian analysis model, so that it can better support the optimization process of serum tumor marker detection.

[0079] Step S104, generating target information according to the prediction weight, the predicted disease progression direction and the associated markers; wherein the target information includes a plurality of target markers.

[0080] Specifically, target information including target markers is generated based on the model output data. The prediction weight and progression direction are comprehensively considered, and the detection of key markers is emphasized. For example, according to the analysis results, it is confirmed that the abnormalities of AFP and CEA are related to the decrease of PTEN protein, and it is recommended to further refine the detection of these proteins. Then the focus of detection is clarified and resource allocation is optimized. By focusing on key markers that may affect disease progression, the clinical significance of detection is enhanced.

[0081] Step S105, controlling the multi-needle detection platform to re-detect serum tumor markers on the preset samples according to the target point information, and completing the array optimization of serum tumor marker detection.

[0082] Specifically, by utilizing the generated target information, the multi-needle detection platform is guided to perform focused detection. The detection parameters and probe settings are reconfigured to improve the accuracy and specificity of the detection. For example, after strengthening the detection of AFP and PTEN proteins, the platform detected weak signal changes based on the recalibrated probes, confirming important markers for early cancer development. Through secondary detection, key signals can be measured more accurately and false positives can be reduced. Dynamic adjustment of the detection process is achieved to improve the overall detection efficiency. The methods provided make serum tumor marker detection faster and more accurate through systematic and data-driven strategies, providing strong support for clinical diagnosis.

[0083] In some embodiments, the preliminary expression information, patient medical history information, imaging information and pathological information are input into a pre-constructed Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker, including: respectively inputting the preliminary expression information, patient medical history information, imaging information and pathological information into corresponding nodes in the Bayesian analysis model, and the Bayesian analysis model calculates the posterior probability distribution information corresponding to each tumor marker in the preliminary expression information based on a preset Bayesian inference algorithm; and obtaining the prediction weight, predicted disease progression direction and correlation marker according to the posterior probability distribution information; wherein the Bayesian inference algorithm includes any one of variational inference and Markov chain Monte Carlo method.

[0084] By inputting the preliminary expression information, patient history information, imaging information and pathology information into the corresponding nodes in the Bayesian model respectively. These information source nodes reflect multidimensional data and match the reasoning structure in the model to ensure that the information can be fully utilized. The Bayesian analysis model uses variational reasoning or Markov chain Monte Carlo (MCMC) method for calculation:

[0085] Variational inference refers to improving computational efficiency by converting inference problems into optimization problems. This method is suitable for processing complex models and large data sets. Markov chain Monte Carlo method (MCMC) is a random sampling method that gradually approaches the true posterior distribution. It is suitable for situations where high accuracy is required, but the amount of computation is large.

[0086] The Bayesian model calculates the posterior probability distribution of each tumor marker in the preliminary expression information through a preset algorithm. This can quantify the performance of each marker under the existing information conditions and estimate its association with the disease. Based on the posterior probability distribution information, the prediction weight, predicted disease progression direction and associated markers of each tumor marker are extracted. The prediction weight helps identify which markers may be particularly important in disease diagnosis, while the disease progression direction provides information about the possibility of worsening or improving the condition.

[0087] Then, by using variational inference or MCMC methods, it is possible to flexibly respond to different computing needs and ensure that reliable and detailed analysis results can be output when processing complex medical data. At the same time, by using the Bayesian model to estimate the posterior probability, doctors can better evaluate the potential clinical significance of tumor markers and make more accurate treatment decisions and risk assessments. The model comprehensively integrates multiple sources of information, providing medical practice with multi-scale analysis capabilities across a single data type. This helps reveal the potential connections between factors, thereby supporting a more complete understanding of pathophysiology. The output probability distribution and various weights allow clinicians to see the confidence of each diagnosis or prediction, making the decision-making process transparent and explainable, and reducing the chance of misjudgment. The algorithmic method used can be adjusted to cope with different data volumes and structures to ensure the adaptability of the implementation plan. Whether it is the addition of new data types or the calculation of larger sample sizes, it can be effectively supported by model expansion.

[0088] In summary, the above embodiments transform various clinical data into action-oriented medical insights through advanced Bayesian reasoning processing technology, greatly enhancing the scientific nature and application potential of tumor marker detection.

[0089] Exemplarily, the obtaining of the prediction weight, the predicted disease progression direction and the associated marker according to the posterior probability distribution information includes: obtaining the prediction weight according to the posterior probability distribution information; forming a weight vector according to a plurality of the prediction weights; obtaining the concentration information corresponding to each tumor marker according to the preliminary expression level corresponding to each tumor marker; obtaining a disease progression threshold according to the weight vector and the plurality of the concentration information; obtaining the predicted disease progression direction by comparing the disease progression threshold with a preset threshold; wherein the predicted disease progression direction includes any one of worsening, maintenance, remission and stability; and obtaining the associated marker in a preset marker database according to the weight vector.

[0090] The prediction weight of each tumor marker is extracted from the calculated posterior probability distribution. The prediction weight reflects the relative importance of each marker in the current analysis. The prediction weights of each tumor marker are integrated into a weight vector, which represents the importance ranking of all markers in the entire model. According to the data of each tumor marker at the preliminary expression level, its specific concentration information is calculated. This information is used to evaluate the actual biological performance of the marker in serum. According to the weight vector and the concentration information of each marker, a comprehensive disease progression threshold is calculated. This threshold is used to evaluate the changing trend of the overall disease state. The calculated disease progression threshold is compared with a preset threshold, and the direction of disease progression is determined according to the comparison results. Possible progression directions include deterioration, maintenance, remission and stability. Using the generated weight vector, other markers related to the current analysis are searched and identified in the preset marker database. These associated markers may provide more extensive diagnostic support or treatment plan guidance.

[0091] By integrating information from multiple markers, the calculated disease progression threshold provides a method to accurately assess the disease state, allowing doctors to better judge the direction of the patient's health. The application of prediction weights allows the markers with the greatest impact to be prioritized, optimizing the model's predictive ability, especially in distinguishing different directions of disease progression. Based on each patient's specific biomarker concentration and its relative weight, the model can provide personalized health assessments and help develop individualized treatment plans. By identifying associated markers, doctors can obtain more potential treatment and management information, which is particularly useful for complex cases or difficult to diagnose situations. The method used to determine the direction of disease progression is data-based, reduces the uncertainty of decision-making, and enhances the scientificity and reliability of clinical decisions. In summary, the above example achieves a fine-grained identification of the role of tumor markers in disease progression through a series of sophisticated calculation and analysis steps, allowing medical practice to formulate more targeted and scientific individualized plans.

[0092] In some embodiments, generating target target information based on the prediction weights, predicted disease progression direction and associated markers includes: determining an abnormal weight among multiple prediction weights; obtaining a weight distribution corresponding to the remaining prediction weights except the abnormal weight; correcting the abnormal weight according to the weight distribution; generating target target information based on the corrected prediction weights, predicted disease progression direction and associated markers.

[0093] Exemplarily, the modifying the abnormal weight according to the weight distribution includes: obtaining the standard deviation, mean and median corresponding to the remaining predicted weights except the abnormal weight according to the weight distribution; modifying the abnormal weight according to the standard deviation, mean and median, and the expression of the abnormal weight includes:

[0094] W'=k1*σ+k2*μ+k3*(ε-W);

[0095] Wherein, W' is the abnormal weight after repair, W is the abnormal weight, σ is the standard deviation, μ is the mean, ε is the median, k1 is the first coefficient, k2 is the second coefficient, and k3 is the third coefficient.

[0096] Obtain statistical characteristics of the prediction weights other than the abnormal weights based on the weight distribution, including standard deviation, mean, and median. These parameters provide comprehensive information about the distribution of weight data. Depending on the specific data set and application requirements, the first, second, and third coefficients can be adjusted to flexibly respond to various data characteristics and abnormal situations. This equation takes into account the deviation of outliers from the median and incorporates the overall distribution characteristics for more effective correction.

[0097] By combining multiple statistics, it provides abnormal correction from different angles, reduces the error that may be caused by a single indicator, and makes the correction result more accurate. The corrected weights can more truly reflect the actual situation of the weight distribution, thereby improving reliability in subsequent analysis and decision-making processes. By reasonably correcting outliers, the integrity of useful data is retained, avoiding information loss caused by direct elimination of outliers, and improving overall data utilization efficiency. The method can provide corrected and optimized input weights for more complex data analysis models, improving the predictive ability and accuracy of the model, especially in medical or bioinformatics research with cross-influence of multiple factors.

[0098] It should be noted that, in some embodiments, the expression of the abnormal weight includes:

[0099]

[0100] Wherein, W' is the abnormal weight after repair, W is the abnormal weight, σ is the standard deviation, μ is the mean, ε is the median, k1 is the first coefficient, k2 is the second coefficient, k3 is the third coefficient, is the interquartile range of the predicted weights except the abnormal weight, which is used to measure the dispersion of the weight distribution and serve as a reference for the compensation mechanism. τ is the quantile of the predicted weights except the abnormal weight, and the position of the abnormal value is adjusted by a specific quantile. Typically, the 25th or 75th percentile can be used to provide a symmetrical measurement standard.

[0101] In some embodiments, generating target point information according to the prediction weight, predicted disease progression direction and associated markers includes: confirming a target weight among a plurality of the prediction weights; determining the tumor marker and the associated marker corresponding to the target weight as suspicious markers; calculating the influencing factors of each predicted disease progression direction and the suspicious marker; determining a target marker among a plurality of the suspicious markers according to the influencing factors, so as to form the target point according to the target marker; wherein the expression of each of the influencing factors is:

[0102] I ij =k j *P(D i |M j )*C ij ;

[0103] Among them, I ij The predicted disease development direction D for the i-th i and the jth suspected marker M j The influencing factors, k j is the weight of the jth target marker, C ij is the correlation coefficient between the i-th suspected marker and the j-th target marker, P(D i |M j ) is used to describe the suspected marker M j The following disease development direction D i The corresponding conditional probability.

[0104] By identifying target weights with special importance or significant characteristics among multiple prediction weights. Target weights represent markers that may have a significant impact on disease progression. Tumor markers and association markers corresponding to target weights are classified as suspected markers. Suspected markers are those that require special attention in the analysis because they may have a significant impact on disease progression. Target markers are selected from multiple suspected markers based on influencing factors. Target markers are those markers that are considered to have the strongest indicative effect on disease progression after comprehensive evaluation. The combination of selected target markers constitutes target target information. This step is an important part of applying the analysis results to clinical research or treatment plan design.

[0105] By identifying highly influential markers, the accuracy and pertinence of disease prediction models are improved, making the results more practical for clinical decision-making. At the same time, comprehensive consideration of multiple influencing factors helps to deeply understand the progression mechanism of tumor-related diseases and the real role of markers in disease development. Target information can help develop personalized treatment plans and conduct precise interventions based on the biological characteristics of specific patients. Confirmed target markers may become new diagnostic and therapeutic targets, shorten the time from research to clinical application, and improve the effectiveness of clinical treatment. Provide potential targets for new drug development and support drug research and development in a more effective and safe direction. Then use mathematical models and data analysis methods to reduce subjective judgment errors and improve the scientific nature of clinical and R&D decisions.

[0106] In summary, the above embodiments successfully convert complex biological data into target information with clinical value through a multi-level and multi-angle analysis process, thereby promoting research and clinical practice in the biomedical field.

[0107] It should be noted that the conditional probability can be calculated in a large database of markers and disease cases in the presence of marker M. j In all cases, D i Frequency, conditional probability calculation can also be performed using Bayesian formula in combination with prior probability and likelihood function, and conditional probability estimation can also be performed using statistical models (such as logistic regression) or machine learning models (such as random forests, support vector machines) by training data. These models can learn the relationship between markers and disease progression during training, and perform probability prediction on new data, and graphical models (such as Bayesian networks or Markov networks) can also be used to build a complex probability graph by nodes (representing markers and disease progression) and edges (representing conditional probability), so as to calculate conditional probability. It is also possible to continuously update the joint probability and conditional probability of markers and disease progression over time in a dynamic environment by Bayesian dynamic update method, reflecting the real-time changes of data, and the present application embodiment is not limited to this.

[0108] The present application provides a method for optimizing a serum tumor marker detection array, which is applied to the control module of the array device provided in the embodiment of the present application. The method detects a preset sample on a microfluidic chip through a multi-needle detection platform to obtain the preliminary expression levels of multiple tumor markers including alpha-fetoprotein and carcinoembryonic antigen. In addition to marker detection, the patient's medical history information, imaging information and pathological information are also combined to form a more comprehensive diagnosis basis. The above information is input into a pre-built Bayesian analysis model. The model outputs the prediction weight of each tumor marker, the predicted direction of disease progression and the correlation marker according to the input data. Based on the results of the Bayesian analysis, a set of optimized target marker lists is generated. The multi-needle detection platform re-tests the sample according to the target target information to optimize the detection strategy. The method and system provided are of great significance for improving the ability of primary hospitals in tumor detection, especially in the case of limited resources, providing an efficient and innovative solution for early diagnosis and monitoring.

[0109] At the same time, the provided method also has the following beneficial effects:

[0110] 1. Improve detection efficiency: A single test can cover more tumor markers, reduce the number of tests, and improve patient compliance.

[0111] 2. Data-driven optimization: With the help of Bayesian analysis models, personalized marker optimization is achieved by integrating patient medical history and test results, which is more targeted and accurate.

[0112] 3. Resource conservation: By optimizing the marker selection and detection process, resources can be effectively utilized, which is particularly suitable for application in primary medical institutions.

[0113] 4. Improve diagnostic accuracy: Integrating multimodal information and multi-marker data can provide more accurate disease assessment and prediction compared to a single method.

[0114] 5. Reduce the burden on patients: Reduce the need for frequent sampling, making the patient's medical treatment process more convenient and comfortable.

[0115] In order to implement the serum tumor marker detection array optimization method corresponding to the above method embodiment, to achieve the corresponding functions and technical effects. Figure 3 , Figure 3 The structural block diagram of a serum tumor marker detection array optimization system 200 provided in an embodiment of the present application is shown. The system is applied to a control module of an array device provided in any embodiment of the present application. For ease of description, only the parts related to the present embodiment are shown. The serum tumor marker detection array optimization system 200 provided in an embodiment of the present application includes:

[0116] The preliminary detection unit 201 is used to obtain preliminary expression information sent by the multi-needle detection platform when the multi-needle detection platform performs serum tumor marker detection on the preset sample; wherein the preliminary expression information includes multiple tumor markers and preliminary expression levels corresponding to each tumor marker, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen;

[0117] An information acquisition unit 202 is used to acquire the patient's medical history information, imaging information, and pathology information corresponding to the preset sample;

[0118] An information input unit 203, used to input the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, wherein the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker;

[0119] A target generating unit 204, configured to generate target information according to the prediction weight, the predicted disease progression direction and the associated marker; wherein the target information includes a plurality of target markers;

[0120] The array optimization unit 205 is used to control the multi-needle detection platform to re-detect the serum tumor markers on the preset sample according to the target point information, so as to complete the array optimization of the serum tumor marker detection.

[0121] The above-mentioned serum tumor marker detection array optimization system 200 can implement the serum tumor marker detection array optimization method of the above-mentioned method embodiment. The options in the above-mentioned method embodiment are also applicable to this embodiment and will not be described in detail here. The rest of the contents of the embodiment of this application can refer to the contents of the above-mentioned method embodiment, and will not be repeated in this embodiment.

[0122] Figure 4 This is a schematic diagram of the structure of a control module provided in one embodiment of the present application. Figure 4 As shown, the control module 13 of this embodiment includes: at least one processor 130 ( Figure 4 Only one is shown in the figure), a memory 131, and a computer program 132 stored in the memory 131 and executable on the at least one processor 130, wherein the processor 130 implements the steps of any of the above method embodiments when executing the computer program 132.

[0123] The control module 13 may be a microcontroller mounted on a smart phone, a tablet computer, a desktop computer, or a cloud server. The control module may include, but is not limited to, a processor 130 and a memory 131. Those skilled in the art will appreciate that Figure 4This is only an example of the control module 13 and does not constitute a limitation on the control module 13 , which may include more or fewer components than shown in the figure, or a combination of certain components, or different components, for example, may also include input and output devices, network access devices, etc.

[0124] The processor 130 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0125] In some embodiments, the memory 131 may be an internal storage unit of the control module 13, such as a hard disk or memory of the control module 13. In other embodiments, the memory 131 may also be an external storage device of the control module 13, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the control module 13. Further, the memory 131 may also include both an internal storage unit of the control module 13 and an external storage device. The memory 131 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 131 may also be used to temporarily store data that has been output or is to be output.

[0126] In addition, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0127] An embodiment of the present application provides a computer program product. When the computer program product runs on a control module, the control module implements the steps in the above-mentioned method embodiments when executed.

[0128] In several embodiments provided in the present application, it is understood that each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved.

[0129] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a control module to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0130] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A method for optimizing a serum tumor marker detection array, characterized in that: The method is applied to a control module of an array device, wherein the array device further comprises a multi-needle detection platform and a microfluidic chip, wherein the multi-needle detection platform is used to simultaneously detect multiple biomarkers on the microfluidic chip, wherein a preset sample is placed in the microfluidic chip; the method comprises: When the multi-needle detection platform performs serum tumor marker detection on a preset sample, obtaining preliminary expression information sent by the multi-needle detection platform; wherein the preliminary expression information includes multiple tumor markers and a preliminary expression level corresponding to each tumor marker, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen; Obtain patient medical history information, imaging information, and pathology information corresponding to the preset sample; The preliminary expression information, patient medical history information, imaging information and pathological information are input into a pre-built Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight corresponding to each tumor marker, the predicted disease progression direction and the correlation marker; Generate target information according to the prediction weight, the predicted disease progression direction and the associated markers; wherein the target information includes multiple target markers; The multi-needle detection platform is controlled to re-test the serum tumor markers of the preset samples according to the target information, and the array optimization of the serum tumor marker detection is completed.

2. The method according to claim 1, characterized in that The Bayesian analysis model includes a plurality of nodes, a directed acyclic graph, and a conditional probability table; before inputting the preliminary expression information, the patient's medical history information, the imaging information, and the pathological information into the pre-built Bayesian analysis model, the method further includes: The preliminary expression information, patient medical history information, imaging information and pathological information are respectively used as variable information of the Bayesian analysis model to be constructed; Constructing a node corresponding to each variable information in the Bayesian analysis model to be constructed; Obtaining dependency information of each of the nodes and the remaining nodes; Generating a directed acyclic graph in the Bayesian analysis model according to the dependency information corresponding to each of the nodes; Acquiring a historical data set corresponding to the serum tumor marker detection; The conditional probability table is generated according to a preset estimation algorithm and the historical data set to complete the construction of the Bayesian analysis model.

3. The method according to claim 1, characterized in that Before inputting the preliminary expression information, patient medical history information, imaging information and pathology information into the pre-built Bayesian analysis model, the method further includes: One-hot encode the patient's medical history information to obtain first feature information; Performing principal component analysis on the pathological information to obtain second characteristic information; Performing image feature extraction on the imaging information to obtain third feature information; The first feature information, the second feature information and the third feature information are normalized; the first feature information, the second feature information and the third feature information are input into the Bayesian analysis model to obtain the prediction weight, the predicted disease progression direction and the correlation marker corresponding to each tumor marker.

4. The method according to claim 1, characterized in that: The preliminary expression information, patient medical history information, imaging information and pathological information are input into a pre-built Bayesian analysis model, and the Bayesian analysis model outputs the prediction weight corresponding to each tumor marker, the predicted disease progression direction and the associated marker, including: The preliminary expression information, patient medical history information, imaging information and pathological information are respectively input into corresponding nodes in the Bayesian analysis model, and the Bayesian analysis model calculates the posterior probability distribution information corresponding to each tumor marker in the preliminary expression information based on a preset Bayesian inference algorithm; Obtaining the prediction weight, predicted disease progression direction and associated markers according to the posterior probability distribution information; Wherein, the Bayesian inference algorithm includes any one of variational inference and Markov chain Monte Carlo method.

5. The method according to claim 4, characterized in that The obtaining of the prediction weight, the prediction direction of disease progression and the associated markers according to the posterior probability distribution information includes: Acquire the prediction weight according to the posterior probability distribution information; Constructing a weight vector according to a plurality of the prediction weights; Acquiring concentration information corresponding to each tumor marker according to the preliminary expression level corresponding to each tumor marker; Acquire a disease progression threshold according to the weight vector and the plurality of concentration information; Comparing the disease progression threshold with a preset threshold to obtain the predicted disease progression direction; wherein the predicted disease progression direction includes any one of deterioration, maintenance, remission and stability; The associated marker is obtained from a preset marker database according to the weight vector.

6. The method according to claim 1, characterized in that The generating target information according to the prediction weight, the predicted disease progression direction and the associated markers includes: determining an anomaly weight among a plurality of said prediction weights; Obtaining weight distribution corresponding to the remaining prediction weights except the abnormal weight; Modifying the abnormal weight according to the weight distribution; The target information is generated based on the revised prediction weight, predicted disease progression direction and associated markers.

7. The method according to claim 6, characterized in that The modifying the abnormal weight according to the weight distribution includes: Obtaining the standard deviation, mean and median corresponding to the remaining prediction weights except the abnormal weight according to the weight distribution; The abnormal weight is modified according to the standard deviation, mean and median, and the expression of the abnormal weight includes: W'=k1*σ+k2*μ+k3*(ε-W); Wherein, W' is the abnormal weight after repair, W is the abnormal weight, σ is the standard deviation, μ is the mean, ε is the median, k1 is the first coefficient, k2 is the second coefficient, and k3 is the third coefficient.

8. The method according to claim 1, characterized in that The generating target information according to the prediction weight, the predicted disease progression direction and the associated markers includes: determining a target weight among the plurality of predicted weights; determining the tumor marker and the associated marker corresponding to the target weight as suspicious markers; Calculating the influencing factors of each of the predicted disease progression directions and the suspected markers; Determining a target marker from the plurality of suspicious markers according to the influencing factors, so as to form the target point according to the target marker; Wherein, the expression of each influencing factor is: I ij =k j *P(D i |M j )*C ij ; Among them, I ij The predicted disease development direction D for the i-th i and the jth suspected marker M j The influencing factors, k j is the weight of the jth target marker, C ij is the correlation coefficient between the i-th suspected marker and the j-th target marker, p(D i |M j ) is used to describe the given suspected marker M j The following disease development direction D i The corresponding conditional probability.

9. An array device, characterized in that: The array device comprises a control module, a multi-needle detection platform and a microfluidic chip, wherein the multi-needle detection platform is used to simultaneously detect multiple biomarkers on the microfluidic chip, and a preset sample is placed in the microfluidic chip; The control module comprises a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the method as claimed in any one of claims 1 to 8 when executing the computer program.

10. A serum tumor marker detection array optimization system, characterized in that: The system is applied to the control module of the array device as claimed in claim 9; the system comprises: A preliminary detection unit, used for obtaining preliminary expression information sent by the multi-needle detection platform when the multi-needle detection platform detects serum tumor markers on the preset sample; wherein the preliminary expression information includes a plurality of tumor markers and a preliminary expression level corresponding to each of the tumor markers, and the tumor markers include at least alpha-fetoprotein and carcinoembryonic antigen; An information acquisition unit, used to acquire the patient's medical history information, imaging information and pathological information corresponding to the preset sample; An information input unit, used to input the preliminary expression information, patient medical history information, imaging information and pathological information into a pre-built Bayesian analysis model, wherein the Bayesian analysis model outputs the prediction weight, predicted disease progression direction and correlation marker corresponding to each tumor marker; A target generating unit, used to generate target information according to the prediction weight, the predicted disease progression direction and the associated marker; wherein the target information includes a plurality of target markers; The array optimization unit is used to control the multi-needle detection platform to re-detect the serum tumor markers on the preset samples according to the target information, so as to complete the array optimization of the serum tumor marker detection.

Images