Prediction model, construction method and prediction system for invasion and metastasis of breast cancer cells

By integrating clinical and genomic data, a deep neural network model was constructed, which solved the problem of insufficient accuracy in predicting invasion and metastasis of breast cancer in the existing technology, and achieved high accuracy and reliability prediction results, which were used to diagnose breast cancer early and Treatment provides strong support.

CN119069135APending Publication Date: 2024-12-03SHENZHEN HOSPITAL OF SOUTHERN MEDICAL UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411001595.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-25
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art relies on clinical pathological characteristics and biomarkers, and has certain limitations and inaccuracies, making it difficult to accurately predict the invasion and metastasis of breast cancer cells.

Method used

By integrating clinical data and genomic data, using bioinformatics and statistical methods to screen features, and building a prediction model based on deep neural networks, combining image preprocessing, feature extraction, model construction, verification and optimization modules to form an intelligent prediction system.

Benefits of technology

It improves the accuracy and reliability of breast cancer cell invasion and metastasis prediction, provides a user-friendly interface, generates intuitive prediction reports, supports clinical decision-making, and provides strong technical support for the early diagnosis and treatment of breast cancer.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119069135A_ABST
    Figure CN119069135A_ABST
Patent Text Reader

Abstract

The invention provides a breast cancer cell invasion and metastasis prediction model, a construction method and a prediction system, and relates to the technical field of breast cancer cell invasion and metastasis prediction. The construction method of the breast cancer cell invasion and metastasis prediction model comprises the following steps: 1, collecting clinical data and genome data of a certain number of breast cancer patients, and preprocessing the data; 2, screening out features closely related to breast cancer invasion and metastasis from the collected data by using a bioinformatics method and a statistics method, and constructing a feature set; 3, constructing a prediction model based on a deep neural network by using the selected feature set; 4, testing the performance of the model on the independent verification set through an evaluation index, and adjusting the structure or parameters of the model according to a verification result until a satisfactory prediction effect is achieved; and 5, embedding the optimized prediction model into a user-friendly interface. The method can more accurately predict invasion and metastasis conditions of breast cancer cells.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of breast cancer cell invasion and metastasis prediction, and in particular to a breast cancer cell invasion and metastasis prediction model, a construction method and a prediction system. Background Art

[0002] Breast MRI is a multi-parameter imaging that can provide morphological and functional information, and therefore plays an important role in the early diagnosis, efficacy evaluation, and prognosis of breast cancer. The diffusion-weighted sequence detects the diffusion of water molecules in different tissues to differentiate between benign and malignant lesions. In normal tissues, water molecules diffuse freely and their diffusion motion is not restricted. Malignant tumors are more densely arranged than benign tumor cells, so the diffusion motion of water molecules is more severely restricted, showing a significant high signal on the DWI sequence. It can reflect the density of tumor cells at the cellular level without the use of contrast agents, and has a high diagnostic value for lesion detection, differentiation of benign and malignant lesions, and assessment of the prognosis of breast tumors, and can improve the specificity of breast MRI diagnosis.

[0003] Breast cancer is one of the most common malignant tumors in women, and its invasion and metastasis are the main causes of death. Traditional diagnostic methods can often only clearly identify metastasis in the late stages of cancer, limiting the effectiveness of treatment. Currently, the prediction of breast cancer cell invasion and metastasis mainly relies on clinical pathological characteristics and biomarkers, but these methods have certain limitations and inaccuracies. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a breast cancer cell invasion and metastasis prediction model and construction method and prediction system, which solves the problem that the prior art relies on clinical pathological characteristics and biomarkers, resulting in certain limitations and inaccuracies.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented by the following technical scheme: A method for constructing a breast cancer cell invasion and metastasis prediction model comprises the following steps:

[0008] Step 1: Data integration and preprocessing:

[0009] Collect clinical data and genomic data of a certain number of breast cancer patients and pre-process the data to ensure data quality and compatibility;

[0010] Step 2: Feature selection:

[0011] Using bioinformatics and statistical methods, we screened out features closely related to breast cancer invasion and metastasis from the collected data and constructed a feature set:

[0012] Step 3: Build the model:

[0013] Using the selected feature set, a prediction model based on deep neural network is constructed;

[0014] Step 4: Verify the model:

[0015] Test the model performance through evaluation indicators on an independent validation set, and adjust the model structure or parameters according to the validation results until a satisfactory prediction effect is achieved;

[0016] Step 5: Optimize the model:

[0017] The optimized prediction model is embedded in a user-friendly interface to form an intelligent prediction system.

[0018] Preferably, in step 1, the clinical data includes clinical pathological information, specifically, tumor size, lymph node status, pathological grade, and ER / PR / HER2 receptor status;

[0019] The genomic data includes gene expression profiles, specifically, RNA sequencing data, mass spectrometry analysis data, MRI, and CT medical imaging data.

[0020] Preferably, in step one, the preprocessing includes data cleaning, standardization and normalization processing.

[0021] Preferably, in step 2, the bioinformatics method includes gene differential expression analysis, gene co-expression network analysis, pathway analysis and survival analysis;

[0022] The statistical methods include random forest and LASSO regression.

[0023] Preferably, in step 2, the characteristics include genes, proteins or other biomarkers.

[0024] Preferably, in step three, the deep neural network includes an Effective Net or Transformer architecture, and cross-validation is used in the model construction process to optimize hyperparameters to ensure the generalization ability of the model.

[0025] Preferably, in step 4, the evaluation indicators include but are not limited to accuracy, sensitivity, specificity and AUC value.

[0026] Preferably, in step five, the prediction system has the ability to receive user input, automatically perform prediction analysis, and generate an intuitive prediction report.

[0027] The present invention further discloses a prediction system of a breast cancer cell invasion and metastasis prediction model, based on the method for constructing the breast cancer cell invasion and metastasis prediction model as described above, comprising:

[0028] Image preprocessing module: used to collect and preprocess data of breast cancer patients;

[0029] Feature extraction module: used to extract features from data and construct sets;

[0030] Model building module: used to build prediction models;

[0031] Model verification module: used to verify the performance of the prediction model;

[0032] Model optimization module: used to optimize the prediction model and form an intelligent prediction system.

[0033] (III) Beneficial effects

[0034] The present invention provides a breast cancer cell invasion and metastasis prediction model and construction method and prediction system, which have the following beneficial effects:

[0035] 1. The breast cancer cell invasion and metastasis prediction model and construction method provided by the present invention can more accurately predict the invasion and metastasis of breast cancer cells by integrating clinical data and genomic data and applying advanced bioinformatics and statistical methods for feature selection. Compared with traditional methods that rely on clinical pathological characteristics and biomarkers, the method of the present invention has higher prediction accuracy and reliability.

[0036] 2. The present invention uses deep neural network technology to build a prediction model, which can automatically learn complex nonlinear relationships from a large amount of data, further improving the performance of the prediction model. At the same time, cross-validation is used in the model construction process to optimize hyperparameters, ensuring the generalization ability of the model and making the prediction results more robust and reliable.

[0037] 3. The prediction system provided by the present invention has a user-friendly interface, can receive user input and automatically perform prediction analysis, and generate intuitive prediction reports. This enables doctors to obtain prediction results conveniently and quickly, providing strong support for clinical decision-making.

[0038] In summary, the breast cancer cell invasion and metastasis prediction model, construction method and prediction system provided by the present invention have the advantages of high accuracy, good reliability, ease of use and strong scalability, and provide strong technical support for the early diagnosis and treatment of breast cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the construction method of the present invention;

[0040] Figure 2 It is a schematic diagram of the prediction system structure of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Embodiment 1:

[0043] like Figure 1 As shown, the embodiment of the present invention provides a method for constructing a breast cancer cell invasion and metastasis prediction model, comprising the following steps:

[0044] Step 1: Data integration and preprocessing:

[0045] First, we collected clinical data and genomic data from a certain number of breast cancer patients. Clinical data mainly include key pathological information such as tumor size, lymph node status, pathological grade, and ER / PR / HER2 receptor status; while genomic data covers RNA sequencing data, mass spectrometry data, and medical imaging data such as MRI and CT. In order to ensure the quality and compatibility of the data, we strictly cleaned, standardized, and normalized the data.

[0046] Step 2: Feature selection:

[0047] Next, we used advanced bioinformatics methods, such as gene differential expression analysis, gene co-expression network analysis, pathway analysis, and survival analysis, to conduct in-depth mining of the collected data. At the same time, combined with statistical methods such as random forests and LASSO regression, we screened out features closely related to breast cancer invasion and metastasis from massive data and constructed a feature set. These features may include specific genes, proteins, or other biomarkers.

[0048] Step 3: Build the model:

[0049] After feature selection, we used the selected feature set to build a prediction model based on a deep neural network. To ensure the performance and generalization ability of the model, we chose advanced neural network architectures such as Effective Net or Transformer, and used cross-validation techniques to optimize hyperparameters during model construction.

[0050] Step 4: Verify the model:

[0051] To evaluate the performance of the model, we tested it on an independent validation set. The model was comprehensively evaluated using multiple evaluation indicators such as accuracy, sensitivity, specificity, and AUC value. Based on the validation results, we continuously adjusted the structure or parameters of the model until a satisfactory prediction effect was achieved.

[0052] Step 5: Optimize the model:

[0053] Finally, we embedded the optimized prediction model into a user-friendly interface to form an intelligent prediction system. This system can receive user input data (such as gene expression data), automatically perform prediction analysis, and generate intuitive prediction reports. This allows doctors to obtain prediction results quickly and easily, providing strong support for clinical decision-making.

[0054] Through the method in Example 1, we successfully built a breast cancer cell invasion and metastasis prediction model and prediction system with high accuracy, good reliability and ease of use. This system not only provides strong technical support for the early diagnosis and treatment of breast cancer, but also demonstrates the strong potential and application value of deep neural network technology in the biomedical field.

[0055] Secondly, if Figure 2 As shown, the embodiment of the present invention also provides a prediction system for the breast cancer cell invasion and metastasis prediction model, including an image preprocessing module, a feature extraction module, a model construction module, a model verification module, and a model optimization module. The functions of each module are as follows:

[0056] Image preprocessing module: This module is responsible for preprocessing the medical imaging data of breast cancer patients. It includes denoising, enhancement, segmentation and other operations to extract key information related to breast cancer invasion and metastasis, and convert this information into a digital format that can be processed by the model. Through this step, the quality and accuracy of the input data can be ensured, laying a solid foundation for subsequent feature extraction and model building.

[0057] Feature extraction module: After data preprocessing is completed, the feature extraction module is responsible for extracting features related to breast cancer invasion and metastasis from the processed data. These features may include gene expression levels, protein abundance, cell morphology, etc. The module uses advanced bioinformatics and machine learning algorithms to automatically mine these features from the data and construct a feature set. Through feature extraction, the data dimension can be greatly reduced, and the computational efficiency and prediction accuracy of the model can be improved.

[0058] Model building module: After feature extraction is completed, the model building module is responsible for building a prediction model using the selected feature set. This module uses deep neural network technology, such as convolutional neural network (CNN) and recurrent neural network (RNN), to automatically learn complex nonlinear relationships in the data. During the model building process, the module also uses cross-validation technology to optimize hyperparameters to ensure the generalization ability and prediction performance of the model.

[0059] Model Validation Module: The model validation module is responsible for evaluating the performance of the constructed prediction model. It uses an independent validation set to test the model and comprehensively evaluates the model through multiple evaluation indicators such as accuracy, sensitivity, specificity, and AUC value. Based on the validation results, the module will give a performance report of the model and guide the model optimization module to adjust and optimize the model.

[0060] Model optimization module: The model optimization module is responsible for optimizing the prediction model based on the feedback results of the model verification module. It may adjust the structure and parameters of the model or introduce a new feature set to improve the prediction accuracy and reliability of the model. The optimized model will be embedded in the prediction system to form an intelligent prediction tool to provide doctors with convenient and fast prediction services.

[0061] Through the collaborative work of the above five modules, the breast cancer cell invasion and metastasis prediction system provided by the present invention can achieve efficient and accurate prediction functions. The system not only provides strong support for the early diagnosis and treatment of breast cancer, but also provides reference and reference for other disease prediction research in the biomedical field.

[0062] Embodiment 2:

[0063] On the basis of the first embodiment, this embodiment further optimizes the construction method and prediction system of the breast cancer cell invasion and metastasis prediction model, especially making innovative improvements in the two key steps of feature selection and model construction.

[0064] First, in terms of feature selection, we introduced a feature learning method based on graph embedding technology. This method can effectively extract features closely related to breast cancer invasion and metastasis from complex biological network data. Specifically, we first constructed a biological network related to breast cancer, including gene expression network, protein interaction network, etc. Then, using graph embedding technology, such as Graph Convolutional Network (GCN) or Graph Attention Network (GAT), the nodes in the network (i.e., genes, proteins, etc.) are mapped to a low-dimensional vector space to obtain the feature representation of each node. These feature representations not only contain the information of the node itself, but also integrate the information of the neighboring nodes in the network, so they can more comprehensively reflect the importance of the node in the biological network. In this way, we screened out features that are more closely related to breast cancer invasion and metastasis, and constructed a more accurate and effective feature set.

[0065] Secondly, in terms of model construction, we adopted a deep learning architecture based on multimodal data fusion. Since the clinical data and genomic data of breast cancer have different modalities and properties, directly inputting them into the same model may lead to information loss or conflict. Therefore, we designed a multimodal data fusion architecture that can process data of different modalities separately and fuse them together through a shared feature fusion layer. Specifically, we designed two sub-networks for clinical data and genomic data respectively, and each sub-network uses advanced deep learning techniques (such as convolutional neural networks, recurrent neural networks, etc.) to extract their respective features. Then, in the feature fusion layer, we concatenate or weightedly fuse the features extracted by the two sub-networks to obtain the final fused features. Finally, we input the fused features into a fully connected layer for classification prediction. In this way, we can not only make full use of the information in data of different modalities, but also improve the prediction accuracy and generalization ability of the model.

[0066] Through the method in Example 2, we have successfully built a more advanced and efficient breast cancer cell invasion and metastasis prediction model and prediction system. This system can not only more accurately predict the risk of breast cancer invasion and metastasis, but also provide doctors with more comprehensive and in-depth analysis results, providing more powerful support for clinical decision-making. At the same time, the system also provides reference and reference for other multimodal data fusion tasks, demonstrating the broad application prospects of deep learning technology in the biomedical field.

[0067] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a breast cancer cell invasion and metastasis prediction model, characterized in that: The following steps are involved: Step 1: Data integration and preprocessing: Collect clinical data and genomic data of a certain number of breast cancer patients and pre-process the data to ensure data quality and compatibility; Step 2: Feature selection: Using bioinformatics and statistical methods, we screened out features closely related to breast cancer invasion and metastasis from the collected data and constructed a feature set: Step 3: Build the model: Using the selected feature set, a prediction model based on deep neural network is constructed; Step 4: Verify the model: Test the model performance through evaluation indicators on an independent validation set, and adjust the model structure or parameters according to the validation results until a satisfactory prediction effect is achieved; Step 5: Optimize the model: The optimized prediction model is embedded in a user-friendly interface to form an intelligent prediction system.

2. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 1, characterized in that: In step 1, the clinical data includes clinical pathological information, specifically, tumor size, lymph node status, pathological grade, and ER / PR / HER2 receptor status; The genomic data includes gene expression profiles, specifically, RNA sequencing data, mass spectrometry analysis data, MRI, and CT medical imaging data.

3. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 2, characterized in that: In step one, the preprocessing includes data cleaning, standardization and normalization.

4. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 1, characterized in that: In step 2, the bioinformatics methods include gene differential expression analysis, gene co-expression network analysis, pathway analysis and survival analysis; The statistical methods include random forest and LASSO regression.

5. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 4, characterized in that: In step 2, the features include genes, proteins or other biomarkers.

6. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 1, characterized in that: In step three, the deep neural network includes an Efficient Net or Transformer architecture, and cross-validation is used in the model construction process to optimize hyperparameters to ensure the generalization ability of the model.

7. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 1, characterized in that: In step 4, the evaluation indicators include but are not limited to accuracy, sensitivity, specificity and AUC value.

8. The method for constructing a breast cancer cell invasion and metastasis prediction model according to claim 1, characterized in that: In step five, the prediction system has the ability to receive user input, automatically perform prediction analysis, and generate intuitive prediction reports.

9. A prediction system for a breast cancer cell invasion and metastasis prediction model, based on the method for constructing a breast cancer cell invasion and metastasis prediction model according to any one of claims 1 to 8, characterized in that: include: Image preprocessing module: used to collect and preprocess data of breast cancer patients; Feature extraction module: used to extract features from data and construct sets; Model building module: used to build prediction models; Model verification module: used to verify the performance of the prediction model; Model optimization module: used to optimize the prediction model and form an intelligent prediction system.

Citation Information

Cited By

  • IDH wild-type glioblastoma typing method and system based on multi-modal data fusion

    CN119475257A

  • Multi-modal data fusion pituitary adenoma invasion behavior characteristic modeling method and system

    CN121075674A