Prediction and evaluation method and system for lung cancer sample

Abstract

The invention relates to the technical field of prediction and evaluation of lung cancer samples, and discloses a prediction and evaluation method of a lung cancer sample, which comprises the following steps: collecting clinical information of a lung cancer patient, acquiring biological data of the lung cancer sample, sorting medical image data, extracting key genes, proteins or pathway characteristics from the biological data, important features related to the lung cancer are screened by using a statistical method or a machine learning algorithm, a proper machine learning or deep learning model is selected according to data characteristics and a prediction target, and the interpretability, stability and calculation complexity of the model are considered. According to the method, tumors with high-risk characteristics, such as tumors with high invasiveness and high metastasis potential, can be found and intervened in time through prediction and evaluation of lung cancer samples, so that the disease progress is delayed, the treatment effect is improved, the treatment effect and disease change can be monitored through regular prediction and evaluation of the lung cancer samples in the treatment process, and the treatment effect is improved. The treatment scheme is adjusted in time, and the maximization of the treatment effect is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of prediction and evaluation of lung cancer samples, and in particular to a prediction and evaluation method and system for lung cancer samples. Background Art

[0002] Lung cancer is a malignant tumor that originates in the lungs, that is, a malignant tumor that originates in the bronchi or alveoli, also known as primary bronchoalveolar carcinoma. There are many ways to classify lung cancer. According to the site of the tumor, it is mainly divided into central lung cancer, peripheral lung cancer and diffuse lung cancer. According to the way the tumor grows, it can be divided into intraductal type, wall-infiltrative type, massive type, spherical type, nodular type and diffuse infiltrative type. According to histopathology, it is mainly divided into small cell carcinoma, squamous cell carcinoma, adenocarcinoma, large cell carcinoma, adenosquamous carcinoma and carcinoid.

[0003] Lung cancer samples are mainly collected through surgery, biopsy, cytological examination, etc., including tissue samples, cytological samples and blood samples. Different types of samples are suitable for different detection purposes and methods. Therefore, a predictive evaluation method for lung cancer samples is needed to prevent lung cancer and intervene in treatment in advance. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for predicting and evaluating lung cancer samples, which has the advantages of improving treatment effects and solves the problems mentioned in the above background technology.

[0006] (II) Technical solution

[0007] In order to achieve the purpose of improving the treatment effect mentioned in the above background technology, the present invention provides the following technical solution: a method and system for predicting and evaluating lung cancer samples, comprising the following steps:

[0008] Step 1: Collect clinical information of lung cancer patients, obtain biological data of lung cancer samples, and organize medical imaging data;

[0009] Step 2: Extract key genes, proteins or pathway features from biological data and use statistical methods or machine learning algorithms to screen important features related to lung cancer;

[0010] Step 3: Select an appropriate machine learning or deep learning model based on data characteristics and prediction objectives, and consider the model's interpretability, stability, and computational complexity;

[0011] Step 4: Use methods such as holdout and cross-validation to evaluate model performance within the training data and verify model performance on an independent test data set to ensure that the model has good generalization ability;

[0012] Step 5: Analyze the important features and weights in the model, explain the basis of the model prediction, and apply the model to the prediction evaluation of actual lung cancer samples;

[0013] Step 6: Collect feedback and result data from the model’s actual application, and iteratively update the model based on new data and research results.

[0014] Preferably, collecting clinical information of patients also includes cleaning and preprocessing of data, extraction and selection of features, partitioning of data, and storage and management of data.

[0015] Ensure that there are no duplicate records in the dataset. For missing values, methods such as filling, interpolation, or deletion can be used to handle them. According to data distribution and domain knowledge, identify and handle outliers to avoid their adverse effects on model training. Divide the cleaned and preprocessed dataset into training set, validation set, and test set. The training set is used for model training, the validation set is used for model parameter adjustment and selection of the optimal model, and the test set is used to evaluate the final performance of the model. When dividing the dataset, it is necessary to ensure the independence and same distribution between the subsets, store the prepared data in an appropriate location, and establish a corresponding data management system so that subsequent analysis and modeling work can easily access and use these data.

[0016] Preferably, feature extraction also includes image feature extraction, biological feature extraction, and clinical feature extraction;

[0017] Feature selection is the process of screening out the most important and effective features for predicting the target from the many extracted features. Commonly used selection methods include filtering methods, wrapping methods, embedded methods, and deep learning-based methods.

[0018] Image feature extraction: Extract the size, shape, location, texture, density and other features of the tumor from medical images such as CT, MRI or PET / CT. Common image feature extraction methods include threshold segmentation, region growing, edge detection, etc., as well as automatic extraction of image features using deep learning technology;

[0019] Biological feature extraction: Extract key features from patients’ biological data such as gene expression data, mutation status, and protein levels. Biological feature extraction usually involves bioinformatics methods and statistical analysis techniques, such as gene expression profiling, mutation detection, and proteomics analysis.

[0020] Clinical feature extraction: extract clinical features such as age, gender, smoking history, family history, and past medical history from the patient’s clinical data;

[0021] Filtering method: Select features based on statistical tests or distance metrics. Statistical test methods such as chi-square test and ANOVA can be used to evaluate the correlation or independence between features and prediction targets, and distance metrics such as mutual information and correlation coefficient can be used to evaluate the redundancy between features.

[0022] Wrapper method: consider feature selection as a search optimization problem, evaluate the performance of the model by constructing different feature subsets, and select the feature subset with the best performance;

[0023] Embedded method: feature selection is automatically completed during model training;

[0024] Methods based on deep learning: Deep learning models have the ability to automatically extract and select features. During the training process, the model can automatically learn and filter out feature representations that are useful for predicting targets.

[0025] Preferably, the model construction includes the selection of variables, the choice of algorithms and the construction of models, and the model is verified through data verification, performance evaluation, and model optimization, and the verification results are evaluated.

[0026] The model was validated using independent cohort data to evaluate the model's predictive performance. The validation cohort design should include patients diagnosed with lung cancer and healthy controls, and the sample size should be large enough to ensure the reliability of the results. The k-fold cross-validation method was used to divide the data set into k subsets, and k-1 subsets were used as training sets each time, and the remaining subset was used as the test set. The average value was taken as the evaluation result after k repetitions. Evaluation indicators such as the area under the ROC curve (AUC value) and the harmonic mean of the accuracy and recall rate (F1 score) were used to measure the classification performance and recognition ability of the model. By analyzing the degree of influence of each feature in the model on the prediction results, the key features were found to provide guidance for subsequent research, and the prediction performance was improved by adjusting the model parameters. Multimodal features were integrated to obtain more comprehensive information, and the model was further optimized in combination with clinical experience and domain knowledge.

[0027] Preferably, the model is constructed through main algorithms and technologies, and the main algorithms are: machine learning algorithm, deep learning algorithm, and then the model is applied. The model can make early diagnosis of patients, make treatment decisions, and perform prognosis evaluation after treatment.

[0028] Machine learning algorithms: such as decision trees, random forests, support vector machines, and neural networks. These algorithms can automatically learn rules and patterns from data and are used to predict and classify unknown data.

[0029] Deep learning technology: It is particularly suitable for processing complex medical imaging data and genomic data, and can extract and classify features of images and sequence data through models such as convolutional neural networks;

[0030] Statistical models: such as the Cox proportional hazards regression model, which is used to analyze survival data and predict the prognosis of lung cancer patients;

[0031] Lung nodule malignancy risk stratification: The prediction model is used to stratify the malignancy risk of lung nodules, assisting clinicians in making more scientific clinical decisions and improving the early diagnosis rate of lung cancer. This helps to detect lung cancer patients at an early stage, so that timely and effective treatment measures can be taken to improve the survival rate and quality of life of patients.

[0032] Personalized medicine: Based on the individual characteristics of patients and the results of prediction models, personalized diagnosis and treatment plans are formulated for patients to achieve precision medicine;

[0033] Treatment plan selection: Prediction models can help clinicians evaluate the effects of different treatment plans and patient prognosis, so as to select the most appropriate treatment plan, which helps to optimize the allocation of medical resources and improve treatment effects and patient satisfaction;

[0034] Efficacy monitoring and adjustment: During the treatment process, by continuously collecting patients' clinical information and biomarker data, using predictive models to monitor treatment effects and timely adjust treatment plans, it helps to ensure the effectiveness and safety of treatment.

[0035] Preferably, the iterative updating of the model includes the innovation of technical methods, the discovery and application of biomarkers, and the accumulation and feedback of clinical data;

[0036] Innovations in technical methods include multi-omics data analysis and machine learning and artificial intelligence;

[0037] The discovery and application of biomarkers include the screening of new biomarkers and the application of biomarker combinations;

[0038] The accumulation and feedback of clinical data include the combination of basic research and clinical application and the feedback loop of clinical application.

[0039] Preferably, it includes a data acquisition module, a data preprocessing module, a feature extraction and selection module, a model construction and training module, a prediction and evaluation module, a result display and report generation module, and a model update and optimization module.

[0040] Preferably, the data collection module is responsible for collecting various data related to lung cancer, including basic information of the patient, clinical examination data, pathological data and genetic data.

[0041] Data preprocessing module: cleans, organizes, standardizes and normalizes the collected raw data to eliminate noise and outliers in the data and improve data quality;

[0042] Feature extraction and selection module: extracts features that are important for lung cancer prediction from preprocessed data, and selects the most representative feature set through feature selection algorithm. These features may include imaging features, gene features, and clinical features.

[0043] Model building and training module: Based on the extracted features, a mathematical model that can accurately predict the risk of lung cancer is constructed. This module usually involves the selection of machine learning algorithms, optimization of model parameters, and model training. Through the training data set, the model can learn the relationship between the features and the risk of lung cancer.

[0044] Prediction and evaluation module: Use the constructed model to predict new lung cancer samples, output the prediction results, and interpret and evaluate the results. This module may include the calculation of prediction probability, determination of classification results, and evaluation of prediction performance;

[0045] Result display and report generation module: The prediction results are presented to users in an intuitive way, such as through graphs, tables or reports. This module is designed to help users better understand the prediction results and provide a scientific basis for clinical decision-making;

[0046] Model update and optimization module: With the continuous addition of new data and the continuous improvement of algorithms, the system needs to regularly update and optimize the prediction model to improve the accuracy and reliability of the prediction.

[0047] Core modules such as data collection, data preprocessing, feature extraction and selection, model construction and training, prediction and evaluation, result presentation and report generation, and model update and optimization work together to achieve accurate prediction and evaluation of lung cancer samples.

[0048] Compared with the prior art, the present invention provides a method and system for predicting and evaluating lung cancer samples, which has the following beneficial effects:

[0049] 1. The present invention can timely discover and intervene in tumors with high-risk characteristics, such as highly invasive tumors and tumors with high metastatic potential, through predictive evaluation of lung cancer samples, thereby delaying disease progression and improving treatment effects. During the treatment process, through regular predictive evaluation of lung cancer samples, the treatment effect and disease changes can be monitored, and the treatment plan can be adjusted in time to ensure the maximization of the treatment effect.

[0050] 2. The present invention, accurate prediction evaluation and treatment decision-making help control tumor growth and metastasis, thereby prolonging the patient's survival, by reducing unnecessary drug side effects and treatment complications, accurate prediction evaluation and treatment decision-making help improve the patient's quality of life.

[0051] 3. Through the predictive evaluation of lung cancer samples, the present invention can help doctors understand key information such as the patient's tumor type, genetic variation, tumor marker levels, etc., so as to develop a more personalized treatment plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0053] 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.

[0054] See also Figure 1 , a predictive evaluation method for lung cancer samples, comprising the following steps:

[0055] Step 1: Collect clinical information of lung cancer patients, obtain biological data of lung cancer samples, and organize medical imaging data;

[0056] Step 2: Extract key genes, proteins or pathway features from biological data and use statistical methods or machine learning algorithms to screen important features related to lung cancer;

[0057] Step 3: Select an appropriate machine learning or deep learning model based on data characteristics and prediction objectives, and consider the model's interpretability, stability, and computational complexity;

[0058] Step 4: Use methods such as holdout and cross-validation to evaluate model performance within the training data and verify model performance on an independent test data set to ensure that the model has good generalization ability;

[0059] Step 5: Analyze the important features and weights in the model, explain the basis of the model prediction, and apply the model to the prediction evaluation of actual lung cancer samples;

[0060] Step 6: Collect feedback and result data from the model’s actual application, and iteratively update the model based on new data and research results.

[0061] Specifically, Figure 1As shown, collecting patient clinical information also includes cleaning and preprocessing of data, extraction and selection of features, partitioning of data, and storage and management of data.

[0062] Through the above technical solutions, we ensure that there are no duplicate records in the data set. For missing values, we can use methods such as filling, interpolation or deletion to process them. According to data distribution and domain knowledge, we can identify and process outliers to avoid their adverse effects on model training. The cleaned and preprocessed data set is divided into training set, validation set and test set. The training set is used for model training, the validation set is used for model parameter adjustment and selection of the optimal model, and the test set is used to evaluate the final performance of the model. When dividing the data set, it is necessary to ensure the independence and same distribution between the subsets, store the prepared data in an appropriate location, and establish a corresponding data management system so that subsequent analysis and modeling work can easily access and use these data.

[0063] Specifically, Figure 1 As shown, feature extraction also includes image feature extraction, biological feature extraction, and clinical feature extraction;

[0064] Feature selection is the process of screening out the most important and effective features for predicting the target from the many extracted features. Commonly used selection methods include filtering methods, wrapping methods, embedded methods, and deep learning-based methods.

[0065] Through the above technical solution, image feature extraction: extract the size, shape, location, texture, density and other features of the tumor from medical images such as CT, MRI or PET / CT. Commonly used image feature extraction methods include threshold segmentation, region growing, edge detection, etc., as well as automatic extraction of image features using deep learning technology;

[0066] Biological feature extraction: Extract key features from patients’ biological data such as gene expression data, mutation status, and protein levels. Biological feature extraction usually involves bioinformatics methods and statistical analysis techniques, such as gene expression profiling, mutation detection, and proteomics analysis.

[0067] Clinical feature extraction: extract clinical features such as age, gender, smoking history, family history, and past medical history from the patient’s clinical data;

[0068] Filtering method: Select features based on statistical tests or distance metrics. Statistical test methods such as chi-square test and ANOVA can be used to evaluate the correlation or independence between features and prediction targets, and distance metrics such as mutual information and correlation coefficient can be used to evaluate the redundancy between features.

[0069] Wrapper method: consider feature selection as a search optimization problem, evaluate the performance of the model by constructing different feature subsets, and select the feature subset with the best performance;

[0070] Embedded method: feature selection is automatically completed during model training;

[0071] Methods based on deep learning: Deep learning models have the ability to automatically extract and select features. During the training process, the model can automatically learn and filter out feature representations that are useful for predicting targets.

[0072] Specifically, Figure 1 As shown, the model construction includes the selection of variables, the choice of algorithms and the construction of models, and the verification of the model through data verification, performance evaluation and model optimization, and the evaluation of the verification results.

[0073] Through the above technical solution, the model is validated using independent cohort data to evaluate the model's predictive performance. The validation cohort design should include patients diagnosed with lung cancer and healthy controls, and the sample size should be large enough to ensure the reliability of the results. The k-fold cross-validation method is used to divide the data set into k subsets, and k-1 subsets are used as training sets each time, and the remaining subset is used as a test set. The average value is taken as the evaluation result after k repetitions. The area under the ROC curve (AUC value), the harmonic mean of the accuracy and recall rate (F1 score) and other evaluation indicators are used to measure the classification performance and recognition ability of the model. By analyzing the degree of influence of each feature in the model on the prediction results, the key features are found to provide guidance for subsequent research, and the prediction performance is improved by adjusting the model parameters. Multimodal features are integrated to obtain more comprehensive information, and the model is further optimized in combination with clinical experience and domain knowledge.

[0074] Specifically, Figure 1 As shown, the model is constructed through main algorithms and technologies, and the main algorithms are: machine learning algorithm, deep learning algorithm, and then the model is applied. The model can diagnose patients early, make treatment decisions, and perform prognosis assessment after treatment.

[0075] Through the above technical solutions, machine learning algorithms: such as decision trees, random forests, support vector machines, and neural networks, these algorithms can automatically learn rules and patterns from data for prediction and classification of unknown data;

[0076] Deep learning technology: It is particularly suitable for processing complex medical imaging data and genomic data, and can extract and classify features of images and sequence data through models such as convolutional neural networks;

[0077] Statistical models: such as the Cox proportional hazards regression model, which is used to analyze survival data and predict the prognosis of lung cancer patients;

[0078] Lung nodule malignancy risk stratification: The prediction model is used to stratify the malignancy risk of lung nodules, assisting clinicians in making more scientific clinical decisions and improving the early diagnosis rate of lung cancer. This helps to detect lung cancer patients at an early stage, so that timely and effective treatment measures can be taken to improve the survival rate and quality of life of patients.

[0079] Personalized medicine: Based on the individual characteristics of patients and the results of prediction models, personalized diagnosis and treatment plans are formulated for patients to achieve precision medicine;

[0080] Treatment plan selection: Prediction models can help clinicians evaluate the effects of different treatment plans and patient prognosis, so as to select the most appropriate treatment plan, which helps to optimize the allocation of medical resources and improve treatment effects and patient satisfaction;

[0081] Efficacy monitoring and adjustment: During the treatment process, by continuously collecting patients' clinical information and biomarker data, using predictive models to monitor treatment effects and timely adjust treatment plans, it helps to ensure the effectiveness and safety of treatment.

[0082] Specifically, Figure 1 As shown, the iterative update of the model includes the innovation of technical methods, the discovery and application of biomarkers, and the accumulation and feedback of clinical data;

[0083] Innovations in technical methods include multi-omics data analysis and machine learning and artificial intelligence;

[0084] The discovery and application of biomarkers include the screening of new biomarkers and the application of biomarker combinations;

[0085] The accumulation and feedback of clinical data include the combination of basic research and clinical application and the feedback loop of clinical application.

[0086] Specifically, Figure 1 As shown, it includes data acquisition module, data preprocessing module, feature extraction and selection module, model construction and training module, prediction and evaluation module, result display and report generation module, and model update and optimization module.

[0087] Specifically, Figure 1 As shown, the data acquisition module is responsible for collecting various data related to lung cancer, including basic information of patients, clinical examination data, pathological data and genetic data.

[0088] Data preprocessing module: cleans, organizes, standardizes and normalizes the collected raw data to eliminate noise and outliers in the data and improve data quality;

[0089] Feature extraction and selection module: extracts features that are important for lung cancer prediction from preprocessed data, and selects the most representative feature set through feature selection algorithm. These features may include imaging features, gene features, and clinical features.

[0090] Model building and training module: Based on the extracted features, a mathematical model that can accurately predict the risk of lung cancer is constructed. This module usually involves the selection of machine learning algorithms, optimization of model parameters, and model training. Through the training data set, the model can learn the relationship between the features and the risk of lung cancer.

[0091] Prediction and evaluation module: Use the constructed model to predict new lung cancer samples, output the prediction results, and interpret and evaluate the results. This module may include the calculation of prediction probability, determination of classification results, and evaluation of prediction performance;

[0092] Result display and report generation module: The prediction results are presented to users in an intuitive way, such as through graphs, tables or reports. This module is designed to help users better understand the prediction results and provide a scientific basis for clinical decision-making;

[0093] Model update and optimization module: With the continuous addition of new data and the continuous improvement of algorithms, the system needs to regularly update and optimize the prediction model to improve the accuracy and reliability of the prediction.

[0094] Through the above technical solutions, core modules such as data collection, data preprocessing, feature extraction and selection, model construction and training, prediction and evaluation, result display and report generation, and model update and optimization work together to achieve accurate prediction and evaluation of lung cancer samples.

[0095] 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 predicting and evaluating lung cancer samples, characterized in that: The following steps are involved: Step 1: Collect clinical information of lung cancer patients, obtain biological data of lung cancer samples, and organize medical imaging data; Step 2: Extract key genes, proteins or pathway features from biological data and use statistical methods or machine learning algorithms to screen important features related to lung cancer; Step 3: Select an appropriate machine learning or deep learning model based on data characteristics and prediction objectives, and consider the model's interpretability, stability, and computational complexity; Step 4: Use methods such as holdout and cross-validation to evaluate model performance within the training data and verify model performance on an independent test data set to ensure that the model has good generalization ability; Step 5: Analyze the important features and weights in the model, explain the basis of the model prediction, and apply the model to the prediction evaluation of actual lung cancer samples; Step 6: Collect feedback and result data from the model’s actual application, and iteratively update the model based on new data and research results.

2. The method for predicting and evaluating lung cancer samples according to claim 1, characterized in that: Collecting patient clinical information also includes data cleaning and preprocessing, feature extraction and selection, data segmentation, and data storage and management.

3. The method for predicting and evaluating lung cancer samples according to claim 1, characterized in that: Feature extraction also includes image feature extraction, biological feature extraction, and clinical feature extraction; Feature selection is the process of screening out the most important and effective features for predicting the target from the many extracted features. Commonly used selection methods include filtering methods, wrapping methods, embedded methods, and deep learning-based methods.

4. The method for predicting and evaluating lung cancer samples according to claim 1, characterized in that: The model construction includes the selection of variables, the choice of algorithms and the construction of models, and the verification of the model through data verification, performance evaluation and model optimization, and the evaluation of the verification results.

5. The method for predicting and evaluating lung cancer samples according to claim 1, characterized in that: The model is constructed through major algorithms and technologies, and the main algorithms are: machine learning algorithm, deep learning algorithm, and then the model is applied. The model can make early diagnosis of patients, make treatment decisions, and conduct prognosis assessment after treatment.

6. The method for predicting and evaluating lung cancer samples according to claim 1, characterized in that: The iterative update of the model includes the innovation of technical methods, the discovery and application of biomarkers, and the accumulation and feedback of clinical data; Innovations in technical methods include multi-omics data analysis and machine learning and artificial intelligence; The discovery and application of biomarkers include the screening of new biomarkers and the application of biomarker combinations; The accumulation and feedback of clinical data include the combination of basic research and clinical application and the feedback loop of clinical application.

7. The prediction and evaluation system for lung cancer samples according to claim 1, characterized in that: It includes data acquisition module, data preprocessing module, feature extraction and selection module, model building and training module, prediction and evaluation module, result display and report generation module, and model update and optimization module.

8. The prediction and evaluation system for lung cancer samples according to claim 7, characterized in that: The data collection module is responsible for collecting various data related to lung cancer, including basic information of patients, clinical examination data, pathological data and genetic data. Data preprocessing module: cleans, organizes, standardizes and normalizes the collected raw data to eliminate noise and outliers in the data and improve data quality; Feature extraction and selection module: extracts features that are important for lung cancer prediction from preprocessed data, and selects the most representative feature set through feature selection algorithm. These features may include imaging features, gene features, and clinical features. Model building and training module: Based on the extracted features, a mathematical model that can accurately predict the risk of lung cancer is constructed. This module usually involves the selection of machine learning algorithms, optimization of model parameters, and model training. Through the training data set, the model can learn the relationship between the features and the risk of lung cancer. Prediction and evaluation module: Use the constructed model to predict new lung cancer samples, output the prediction results, and interpret and evaluate the results. This module may include the calculation of prediction probability, determination of classification results, and evaluation of prediction performance; Result display and report generation module: The prediction results are presented to users in an intuitive way, such as through graphs, tables or reports. This module is designed to help users better understand the prediction results and provide a scientific basis for clinical decision-making; Model update and optimization module: With the continuous addition of new data and the continuous improvement of algorithms, the system needs to regularly update and optimize the prediction model to improve the accuracy and reliability of the prediction.

Images