Lung cancer patient prognosis evaluation method and system in combination with pathological image and clinical variable

Abstract

The invention discloses a lung cancer patient prognosis evaluation method and system combining a pathological image and clinical variables, and the method comprises the steps: obtaining the pathological image of a target patient, carrying out the preprocessing and feature extraction, and integrating a plurality of clinical variables according to the identity information and daily habit information of the target patient; constructing a clinical feature network and an image feature network, and analyzing the extracted features through the image feature network to obtain a first analysis result; performing standardization and classification processing on the multiple clinical variables, and analyzing the processed multiple clinical variables through a clinical feature network to obtain a second analysis result; according to the first analysis result and the second analysis result, the survival month of the target patient is predicted through a multi-layer perception model. The lifetime prediction can be comprehensively carried out by considering the cooperative influence of clinical variables such as the tumor microenvironment and the patient smoking history, and the prediction precision and reliability can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cancer prognosis evaluation, and in particular to a lung cancer patient prognosis evaluation method combining pathological images and clinical variables. BACKGROUND

[0002] At present, lung cancer is the main cause of cancer-related deaths worldwide, and non-small cell lung cancer (NSCLC) accounts for 85% of all lung cancer cases. Although surgical resection is the main treatment for early lung cancer, the prognosis of patients with the same stage after surgery varies significantly, and the existing TNM staging system is difficult to accurately predict individual survival outcomes. There is an urgent need for more accurate prognosis evaluation methods to guide personalized treatment and follow-up strategies. At present, lung cancer prognosis evaluation mainly relies on the following methods: observing and evaluating tumor morphology through artificial microscopes, extracting macroscopic tumor features, and analyzing cell morphology through pathological histology models. The lack of systematic integration with clinical variables (such as age, stage, and gene mutations) leads to low prediction accuracy, reducing reliability and stability. SUMMARY

[0003] In view of the above problems, the present application provides a lung cancer patient prognosis evaluation method combining pathological images and clinical variables and a system to solve the problems mentioned in the background art, such as the lack of systematic integration with clinical variables (such as age, stage, and gene mutations) leading to low prediction accuracy, reducing reliability and stability.

[0004] A lung cancer patient prognosis evaluation method combining pathological images and clinical variables, comprising the following steps:

[0005] Obtaining the pathological images of the target patient and performing preprocessing and feature extraction, and integrating multiple clinical variables according to the identity information and daily habit information of the target patient;

[0006] Constructing a clinical feature network and an image feature network, analyzing the extracted features through the image feature network, and obtaining a first analysis result;

[0007] Standardizing and classifying the multiple clinical variables, analyzing the processed multiple clinical variables through the clinical feature network, and obtaining a second analysis result;

[0008] According to the first analysis result and the second analysis result, predicting the survival months of the target patient through a multilayer perception model.

[0009] Preferably, the step of obtaining the pathological images of the target patient and performing preprocessing and feature extraction, and integrating multiple clinical variables according to the identity information and daily habit information of the target patient comprises:

[0010] retrieve lung cancer pathological images of the target patient from the pathological database through the medical record identity of the target patient, and perform image denoising and image enhancement preprocessing on the lung cancer case images;

[0011] perform unsupervised feature learning on the preprocessed lung cancer pathological images and filter high-risk areas, and extract deep semantic features of the high-risk areas through a UNI network;

[0012] determine the age and gender parameters of the target patient according to the identity information of the target patient, and determine the smoking history parameter and the lung cancer triggering correlation parameter of the target patient based on the daily habit information of the target patient;

[0013] integrate the age and gender parameters and the smoking history parameter and the lung cancer triggering correlation parameter of the target patient to generate multiple clinical variables.

[0014] Preferably, the clinical feature network and the image feature network are constructed, the extracted features are analyzed through the image feature network to obtain a first analysis result, including:

[0015] collect clinical data corresponding to each clinical variable, standardize and normalize the clinical data, and determine multiple clinical features and the correlation between the clinical features according to the processed clinical data;

[0016] configure network nodes according to the multiple clinical features and the correlation between the clinical features, and construct a clinical feature network according to the configuration result;

[0017] extract high-level features from pathological image samples through a deep learning model and perform training to construct an image feature network;

[0018] analyze the extracted features through the image feature network to determine the tumor heterogeneity expression ability, and determine the tumor development stage analysis result of the target patient according to the tumor heterogeneity expression ability.

[0019] Preferably, the multiple clinical variables are standardized and classified, and the processed multiple clinical variables are analyzed through the clinical feature network to obtain a second analysis result, including:

[0020] determine the dimension unit of each clinical variable, classify the multiple clinical variables according to the dimension unit, and perform one-hot encoding standardization processing on each type of clinical variable;

[0021] determine the network type of the clinical feature network, determine the analysis strategy according to the network type, and the network type includes a patient network and a variable network;

[0022] analyze the processed each type of clinical variable according to the analysis strategy to determine the subpopulation and disease typing to which the target patient belongs;

[0023] generating a second analysis result based on the subpopulation to which the target patient belongs and the disease type.

[0024] Preferably, the survival month of the target patient is predicted by a multi-layer perception model according to the first analysis result and the second analysis result, comprising:

[0025] determining the target stage of tumor development of the target patient according to the first analysis result, and obtaining the target variable of lymphocytes at the target stage of tumor development;

[0026] determining the target type of lung cancer of the target patient according to the second analysis result, and determining the deterioration state parameter of the target type of lung cancer;

[0027] predicting the survival cycle time of the target patient based on the target variable of lymphocytes at the target stage of tumor development and the deterioration state parameter of the target type of lung cancer by a multi-layer perception model;

[0028] determining the survival month of the target patient based on the survival cycle time and the current time point.

[0029] A lung cancer patient prognosis evaluation system combining pathological images and clinical variables, comprising:

[0030] an acquisition module configured to acquire pathological images of a target patient, perform preprocessing and feature extraction, and integrate multiple clinical variables according to identity information and daily habit information of the target patient;

[0031] a first analysis module configured to construct a clinical feature network and an image feature network, analyze the extracted features by the image feature network, and obtain a first analysis result;

[0032] a second analysis module configured to standardize and classify the multiple clinical variables, analyze the processed multiple clinical variables by the clinical feature network, and obtain a second analysis result;

[0033] a prediction module configured to predict the survival month of the target patient by a multi-layer perception model according to the first analysis result and the second analysis result.

[0034] Preferably, the acquisition module comprises:

[0035] a calling sub-module configured to call lung cancer pathological images of a target patient from a pathological database through the medical record identity of the target patient, and perform image denoising and image enhancement preprocessing on the lung cancer case images;

[0036] an extraction sub-module configured to perform unsupervised feature learning on the preprocessed lung cancer pathological images and screen high-risk areas, and extract deep semantic features of the high-risk areas by a UNI network;

[0037] The first determining sub-module is configured to determine an age and gender parameter of the target patient according to identity information of the target patient, and determine a smoking history parameter and a lung cancer triggering correlation parameter of the target patient based on daily habit information of the target patient.

[0038] The integrating sub-module is configured to integrate the age and gender parameter and the smoking history parameter of the target patient with the lung cancer triggering correlation parameter to generate a plurality of clinical variables.

[0039] Preferably, the first analysis module comprises:

[0040] The second determining sub-module is configured to collect clinical data corresponding to each clinical variable, perform standardization and normalization processing on the clinical data, and determine a plurality of clinical features and correlations between the clinical features according to the processed clinical data.

[0041] The first constructing sub-module is configured to configure network nodes according to the plurality of clinical features and the correlations between the clinical features, and construct a clinical feature network according to a configuration result.

[0042] The second constructing sub-module is configured to extract high-level features from the pathological image sample through a deep learning model and perform training, and construct an image feature network.

[0043] The third determining sub-module is configured to analyze the extracted features through the image feature network to determine a tumor heterogeneity expression capability, and determine a tumor development stage analysis result of the target patient according to the tumor heterogeneity expression capability.

[0044] Preferably, the second analysis module comprises:

[0045] The processing sub-module is configured to determine a dimension unit of each clinical variable, perform classification processing on the plurality of clinical variables according to the dimension unit, and perform one-hot encoding standardization processing on each type of clinical variable.

[0046] The fourth determining sub-module is configured to determine a network type of the clinical feature network, determine an analysis strategy according to the network type, and the network type comprises a patient network and a variable network.

[0047] The fifth determining sub-module is configured to analyze each type of processed clinical variable according to the analysis strategy to determine a sub-population to which the target patient belongs and a disease type.

[0048] The generating sub-module is configured to generate a second analysis result based on the sub-population to which the target patient belongs and the disease type.

[0049] Preferably, the prediction module comprises:

[0050] The sixth determining sub-module is configured to determine a tumor development target stage of the target patient according to the first analysis result, and obtain a lymphocyte target variable at the tumor development target stage.

[0051] a seventh determining sub-module, configured to determine a target type of lung cancer of the target patient according to the second analysis result, and determine a deterioration state parameter of the target type of lung cancer;

[0052] a predicting sub-module, configured to predict a survival cycle time of the target patient by a multilayer perception model based on the target variable of lymphocytes at the target stage of tumor development and the deterioration state parameter of the target type of lung cancer;

[0053] an eighth determining sub-module, configured to determine a survival month of the target patient based on the survival cycle time and a current time point.

[0054] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and attained by the structure particularly pointed out in the written description and claims hereof.

[0055] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0056] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate embodiments of the present application and explain the present application, but do not constitute a limitation of the present application.

[0057] Figure 1 A workflow diagram of a lung cancer patient prognosis evaluation method combining pathological images and clinical variables provided by the present application;

[0058] Figure 2 Another workflow diagram of a lung cancer patient prognosis evaluation method combining pathological images and clinical variables provided by the present application;

[0059] Figure 3 A structural schematic diagram of a lung cancer patient prognosis evaluation system combining pathological images and clinical variables provided by the present application;

[0060] Figure 4 A structural schematic diagram of an acquisition module in a lung cancer patient prognosis evaluation system combining pathological images and clinical variables provided by the present application. DETAILED DESCRIPTION

[0061] The exemplary embodiments will be described in detail below with reference to the drawings. In the following description, the same numbers are used to denote the same elements throughout the several views. The embodiments described in the following exemplary embodiments do not represent all the implementations consistent with the present disclosure. Instead, they simply represent some of the many ways of implementing an apparatus and method consistent with the principles of the present disclosure as detailed in the appended claims.

[0062] Currently, lung cancer is the leading cause of cancer-related deaths worldwide, with non-small cell lung cancer (NSCLC) accounting for 85% of all lung cancer cases. Although surgical resection is the main treatment for early lung cancer, the prognosis of patients with the same stage after surgery varies significantly, and the existing TNM staging system is difficult to accurately predict individual survival outcomes. There is an urgent need for more accurate prognostic evaluation methods to guide personalized treatment and follow-up strategies. Currently, lung cancer prognosis evaluation mainly relies on the following methods: observing tumor morphology through artificial microscope, extracting macroscopic tumor features, and analyzing cell morphology through pathological histology model. The lack of systematic integration with clinical variables (such as age, stage, and gene mutation) results in low prediction accuracy, reducing reliability and stability. To solve the above problems, the present embodiment discloses a lung cancer patient prognosis evaluation method combining pathological images and clinical variables.

[0063] A lung cancer patient prognosis evaluation method combining pathological images and clinical variables, as shown in Figure 1 includes the following steps:

[0064] Step S101, obtaining the pathological image of the target patient and performing preprocessing and feature extraction, integrating multiple clinical variables according to the identity information and daily habit information of the target patient;

[0065] Step S102, constructing a clinical feature network and an image feature network, analyzing the extracted features through the image feature network to obtain a first analysis result;

[0066] Step S103, standardizing and classifying the multiple clinical variables, and analyzing the processed multiple clinical variables through the clinical feature network to obtain a second analysis result;

[0067] Step S104, predicting the survival months of the target patient through a multi-layer perception model according to the first analysis result and the second analysis result.

[0068] The working principle of the technical solution is as follows: a pathological image of a target patient is acquired and preprocessed and feature extraction is performed, a plurality of clinical variables are integrated according to identity information and daily habit information of the target patient; a clinical feature network and an image feature network are constructed, the extracted features are analyzed through the image feature network to obtain a first analysis result; the plurality of clinical variables are standardized and classified, and the processed plurality of clinical variables are analyzed through the clinical feature network to obtain a second analysis result; and the survival month of the target patient is predicted through a multilayer perception model according to the first analysis result and the second analysis result.

[0069] The beneficial effects of the technical solution are as follows: through intelligent combination of pathological image analysis and clinical variable analysis of the patient, the survival period can be comprehensively predicted considering the synergistic influence of the tumor microenvironment and the clinical variables such as the smoking history of the patient, the prediction accuracy and reliability can be ensured, the objectivity and stability of the prediction result are also ensured, and the problem of low prediction result accuracy caused by lack of system integration with clinical variables (such as age, stage, and gene mutation) in the prior art is solved, which reduces the reliability and stability.

[0070] In one embodiment, as shown in Figure 2 The acquiring of the pathological image of the target patient and the pre-processing and feature extraction, and the integrating of the plurality of clinical variables according to the identity information and the daily habit information of the target patient, include:

[0071] Step S201: The lung cancer pathological image of the target patient is retrieved from the pathological database through the medical record identity of the target patient, and the lung cancer case image is preprocessed through image denoising and image enhancement;

[0072] Step S202: Unsupervised feature learning is performed on the preprocessed lung cancer pathological image, and a high-risk area is screened, and deep semantic features of the high-risk area are extracted through a UNI network;

[0073] Step S203: The age and gender parameters of the target patient are determined according to the identity information of the target patient, and the smoking history parameter and the lung cancer triggering correlation parameter of the target patient are determined based on the daily habit information of the target patient;

[0074] Step S204: The age and gender parameters and the smoking history parameter and the lung cancer triggering correlation parameter of the target patient are integrated to generate a plurality of clinical variables.

[0075] The beneficial effects of the above technical solutions are: through deep analysis and feature analysis, high-value pathological features of pathological images can be extracted as reference samples, ensuring the high quality and reliability of the features. Furthermore, by obtaining lung cancer triggering correlation parameters of the target patient as clinical variables, direct and indirect lung cancer pathogenic reasons can be used as reference clinical variables for prognosis analysis, improving the reliability of the analysis results.

[0076] In one embodiment, the constructing a clinical feature network and an image feature network, analyzing the extracted features through the image feature network, and obtaining a first analysis result, includes:

[0077] Collecting clinical data corresponding to each clinical variable, standardizing and normalizing the clinical data, and determining a plurality of clinical features and correlations between the clinical features according to the processed clinical data;

[0078] Configuring network nodes according to the plurality of clinical features and the correlations between the clinical features, and constructing a clinical feature network according to the configuration result;

[0079] Extracting high-level features from pathological image samples through a deep learning model and training the high-level features to construct an image feature network;

[0080] Analyzing the extracted features through the image feature network to determine tumor heterogeneity expression ability, and determining a tumor development stage analysis result of the target patient according to the tumor heterogeneity expression ability.

[0081] The beneficial effects of the above technical solutions are: through the configuration of network nodes, multiple clinical features can be efficiently analyzed, improving the stability.

[0082] In one embodiment, the standardizing and classifying a plurality of clinical variables, analyzing the processed plurality of clinical variables through a clinical feature network, and obtaining a second analysis result, includes:

[0083] Determining the dimension unit of each clinical variable, classifying the plurality of clinical variables according to the dimension unit, and standardizing each type of clinical variable through one-hot encoding;

[0084] Determining the network type of the clinical feature network, determining the analysis strategy according to the network type, and the network type including: a patient network and a variable network;

[0085] Analyzing the processed each type of clinical variable according to the analysis strategy to determine the subpopulation and disease typing to which the target patient belongs;

[0086] Generating a second analysis result based on the subpopulation and disease typing to which the target patient belongs.

[0087] The beneficial effects of the above technical solutions are that the lung cancer disease analysis can be adaptively performed in different ways by combining the analysis target with the analysis strategy, thereby improving the analysis efficiency and stability.

[0088] In one embodiment, the survival month of the target patient is predicted by the multi-layer perception model according to the first analysis result and the second analysis result, including:

[0089] The target stage of tumor development of the target patient is determined according to the first analysis result, and the target variable of lymphocytes at the target stage of tumor development is obtained;

[0090] The target type of lung cancer of the target patient is determined according to the second analysis result, and the deterioration state parameter of the target type of lung cancer is determined;

[0091] The survival cycle time of the target patient is predicted by the multi-layer perception model based on the target variable of lymphocytes at the target stage of tumor development and the deterioration state parameter of the target type of lung cancer;

[0092] The survival month of the target patient is determined based on the survival cycle time and the current time point.

[0093] The beneficial effects of the above technical solutions are that the survival month of the patient can be accurately and intuitively evaluated based on the development characteristics of lung cancer disease by combining the target variable of lymphocytes at the target stage of tumor development and the deterioration state parameter of the target type of lung cancer of the patient, thereby ensuring the reliability of the evaluation result.

[0094] In one embodiment, the present embodiment also discloses a lung cancer patient prognosis evaluation system combining pathological images and clinical variables, as shown in Figure 3 The system includes:

[0095] The acquisition module 301 is configured to acquire the pathological image of the target patient, perform preprocessing and feature extraction, integrate multiple clinical variables according to the identity information and daily habit information of the target patient, and output the first analysis result and the second analysis result.

[0096] The first analysis module 302 is configured to construct a clinical feature network and an image feature network, analyze the extracted features by the image feature network, and obtain the first analysis result.

[0097] The second analysis module 303 is configured to standardize and classify the multiple clinical variables, analyze the processed multiple clinical variables by the clinical feature network, and obtain the second analysis result.

[0098] The prediction module 304 is configured to predict the survival month of the target patient by the multi-layer perception model according to the first analysis result and the second analysis result.

[0099] The working principle and beneficial effects of the technical solutions described above have been explained in the method embodiments, and will not be repeated here.

[0100] In one embodiment, as shown in Figure 4 The acquisition module 301 comprises:

[0101] The calling sub-module 3011 is configured to call the lung cancer pathological image of the target patient from the pathological database by the medical record identity of the target patient, and perform image denoising and image enhancement preprocessing on the lung cancer case image.

[0102] The extraction sub-module 3012 is configured to perform unsupervised feature learning on the preprocessed lung cancer pathological image and filter a high-risk area, and extract deep semantic features of the high-risk area through a UNI network.

[0103] The first determination sub-module 3013 is configured to determine the age and gender parameters of the target patient according to the identity information of the target patient, and determine the smoking history parameter and the lung cancer triggering correlation parameter of the target patient based on the daily habit information of the target patient.

[0104] The integration sub-module 3014 is configured to integrate the age and gender parameters and the smoking history parameter and the lung cancer triggering correlation parameter of the target patient to generate multiple clinical variables.

[0105] In one embodiment, the first analysis module comprises:

[0106] The second determination sub-module is configured to collect clinical data corresponding to each clinical variable, perform standardization and normalization processing on the clinical data, and determine multiple clinical features and the correlation between the clinical features according to the processed clinical data.

[0107] The first construction sub-module is configured to configure network nodes according to the multiple clinical features and the correlation between the clinical features, and construct a clinical feature network according to the configuration result.

[0108] The second construction sub-module is configured to extract high-level features from pathological image samples through a deep learning model and perform training, and construct an image feature network.

[0109] The third determination sub-module is configured to analyze the extracted features through the image feature network to determine the tumor heterogeneity expression ability, and determine the tumor development stage analysis result of the target patient according to the tumor heterogeneity expression ability.

[0110] In one embodiment, the second analysis module comprises:

[0111] The processing sub-module is configured to determine the dimension unit of each clinical variable, perform classification processing on the multiple clinical variables according to the dimension unit, and perform one-hot encoding standardization processing on each type of clinical variable.

[0112] a fourth determining sub-module, configured to determine a network type of the clinical feature network, and determine an analysis strategy according to the network type, wherein the network type comprises a patient network and a variable network;

[0113] a fifth determining sub-module, configured to analyze the processed clinical variables according to the analysis strategy to determine a subpopulation to which the target patient belongs and a disease type of the target patient;

[0114] a generating sub-module, configured to generate a second analysis result based on the subpopulation to which the target patient belongs and the disease type of the target patient.

[0115] In one embodiment, the predicting module comprises:

[0116] a sixth determining sub-module, configured to determine a target tumor development stage of the target patient according to the first analysis result, and acquire a target lymphocyte variable at the target tumor development stage;

[0117] a seventh determining sub-module, configured to determine a target lung cancer type of the target patient according to the second analysis result, and determine a deterioration state parameter of the target lung cancer type;

[0118] a predicting sub-module, configured to predict a survival cycle time of the target patient by a multilayer perception model based on the target lymphocyte variable at the target tumor development stage and the deterioration state parameter of the target lung cancer type;

[0119] an eighth determining sub-module, configured to determine a survival month of the target patient based on the survival cycle time and a current time point.

[0120] It should be understood by those skilled in the art that the first and second in the present disclosure refer to different application stages.

[0121] Other embodiments of the present disclosure will be apparent to those skilled in the art with the consideration of the specification and practice of the disclosure disclosed herein. The present application is intended to cover any variations, uses or adaptive changes of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the art that are not disclosed by the present disclosure. The specification and examples are only regarded as exemplary, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0122] It should be understood that the present disclosure is not limited to the precise construction that has been described and shown in the accompanying drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is only limited by the appended claims.

Claims

1. A method for prognostic assessment of lung cancer patients combining pathological images and clinical variables, characterized in that, Includes the following steps: Pathological images of the target patient were acquired, preprocessed, and feature extracted. Multiple clinical variables were integrated based on the target patient's identity information and daily habits. A clinical feature network and an image feature network are constructed. The extracted features are analyzed through the image feature network to obtain the first analysis result. Multiple clinical variables are standardized and classified, and the processed multiple clinical variables are analyzed through a clinical feature network to obtain the second analysis results; Based on the results of the first and second analyses, the survival months of the target patients are predicted using a multilayer perception model.

2. The method for prognostic assessment of lung cancer patients combining pathological images and clinical variables according to claim 1, characterized in that, The process involves acquiring and preprocessing the pathological images of the target patient, extracting features, and integrating multiple clinical variables based on the target patient's identity and daily habits, including: Lung cancer pathological images of the target patient are retrieved from the pathology database based on the target patient's medical record identity, and the lung cancer case images are preprocessed by image denoising and image enhancement. Unsupervised feature learning was performed on the preprocessed lung cancer pathology images to screen high-risk regions, and deep semantic features of the high-risk regions were extracted through the UNI network. The age and gender parameters of the target patients are determined based on their identity information, and the smoking history parameters and lung cancer provocation correlation parameters of the target patients are determined based on their daily habits information. The age, gender, and smoking history parameters of the target patients were integrated with lung cancer provocation-related parameters to generate multiple clinical variables.

3. The method for prognostic assessment of lung cancer patients combining pathological images and clinical variables according to claim 1, characterized in that, The construction of the clinical feature network and the image feature network, and the analysis of the extracted features through the image feature network to obtain the first analysis result, includes: Collect clinical data corresponding to each clinical variable, standardize and normalize the clinical data, and determine multiple clinical features and the correlation between each clinical feature based on the processed clinical data. Configure network nodes based on multiple clinical features and the correlations between them, and construct a clinical feature network based on the configuration results; High-level features are extracted from pathological image samples using a deep learning model and trained to construct an image feature network. The extracted features are analyzed using an image feature network to determine the tumor heterogeneity expression capacity, and the tumor development stage of the target patient is determined based on the tumor heterogeneity expression capacity.

4. The method for prognostic assessment of lung cancer patients combining pathological images and clinical variables according to claim 1, characterized in that, The standardization and classification of multiple clinical variables, followed by analysis of the processed clinical variables using a clinical feature network to obtain a second analysis result, includes: Determine the dimensional unit for each clinical variable, classify multiple clinical variables according to the dimensional unit, and standardize each type of clinical variable by one-hot coding. Determine the network type of the clinical feature network, and determine the analysis strategy based on the network type. The network types include: patient network and variable network. Based on the analysis strategy, the processed clinical variables are analyzed to determine the target patient's subgroup and disease type; A second analysis result is generated based on the target patient's subgroup and disease type.

5. The method for prognostic assessment of lung cancer patients combining pathological images and clinical variables according to claim 1, characterized in that, The prediction of the target patient's survival months using a multilayer perception model based on the results of the first and second analyses includes: Based on the results of the first analysis, the target stage of tumor development in the target patients was determined, and the target lymphocyte variables at the target stage of tumor development were obtained. Based on the results of the second analysis, the target type of lung cancer in the target patients was determined, and the deterioration status parameters of the target type of lung cancer were determined. The survival time of target patients was predicted by using a multilayer perception model based on lymphocyte target variables at the target stage of tumor development and the deterioration status parameters of the target type of lung cancer. The survival months of the target patient are determined based on the survival period and the current time point.

6. A prognostic assessment system for lung cancer patients that combines pathological images with clinical variables, characterized in that, The system includes: The acquisition module is used to acquire pathological images of the target patient and perform preprocessing and feature extraction, integrating multiple clinical variables based on the target patient's identity information and daily habit information; The first analysis module is used to construct a clinical feature network and an image feature network. The extracted features are analyzed through the image feature network to obtain the first analysis result. The second analysis module is used to standardize and classify multiple clinical variables. It analyzes the processed multiple clinical variables through a clinical feature network to obtain the second analysis results. The prediction module is used to predict the survival months of the target patient based on the results of the first and second analyses using a multilayer perceptual model.

7. The lung cancer patient prognostic assessment system combining pathological images and clinical variables according to claim 6, characterized in that, The acquisition module includes: The retrieval submodule is used to retrieve lung cancer pathology images of the target patient from the pathology database based on the target patient's medical record identity, and to perform image denoising and image enhancement preprocessing on the lung cancer case images. The extraction submodule is used to perform unsupervised feature learning on the preprocessed lung cancer pathology image and screen high-risk areas. The deep semantic features of the high-risk areas are extracted through the UNI network. The first determination submodule is used to determine the age and gender parameters of the target patient based on the target patient's identity information, and to determine the smoking history parameters and lung cancer provocation correlation parameters of the target patient based on the target patient's daily habit information. The integration submodule is used to integrate the target patient's age, gender, smoking history, and lung cancer provocation-related parameters to generate multiple clinical variables.

8. The prognostic assessment system for lung cancer patients combining pathological images and clinical variables according to claim 6, characterized in that, The first analysis module includes: The second determination submodule is used to collect clinical data corresponding to each clinical variable, standardize and normalize the clinical data, and determine multiple clinical features and the correlation between each clinical feature based on the processed clinical data. The first construction submodule is used to configure network nodes based on multiple clinical features and the correlations between clinical features, and to construct a clinical feature network based on the configuration results. The second construction submodule is used to extract high-level features from pathological image samples through a deep learning model and train it to build an image feature network. The third determination submodule is used to analyze the extracted features through an image feature network to determine the tumor heterogeneous expression capacity, and to determine the tumor development stage of the target patient based on the tumor heterogeneous expression capacity.

9. The prognostic assessment system for lung cancer patients combining pathological images and clinical variables according to claim 6, characterized in that, The second analysis module includes: The processing submodule is used to determine the dimensional unit of each clinical variable, classify multiple clinical variables according to the dimensional unit, and perform one-hot coding standardization on each type of clinical variable. The fourth determination submodule is used to determine the network type of the clinical feature network and determine the analysis strategy based on the network type. The network types include: patient network and variable network. The fifth submodule is used to analyze the processed clinical variables according to the analysis strategy to determine the target patient's subgroup and disease type; The generation submodule is used to generate second analysis results based on the target patient's subgroup and disease type.

10. The lung cancer patient prognostic assessment system combining pathological images and clinical variables according to claim 6, characterized in that, The prediction module includes: The sixth determination submodule is used to determine the target stage of tumor development in the target patient based on the results of the first analysis, and to obtain the target lymphocyte variable at the target stage of tumor development. The seventh determination submodule is used to determine the target type of lung cancer in the target patient based on the results of the second analysis, and to determine the deterioration status parameters of the target type of lung cancer; The prediction submodule is used to predict the survival time of target patients based on lymphocyte target variables at the target stage of tumor development and the deterioration status parameters of the target type of lung cancer using a multilayer perceptual model. The eighth determination submodule is used to determine the survival months of the target patient based on the survival cycle time and the current time point.

Images