Tumor disease assessment method based on big data

Abstract

The invention discloses a tumor disease assessment method based on big data, and particularly relates to the technical field of tumor disease assessment. Medical data and living habit data of a patient are collected through multiple data interfaces, integration is carried out through preprocessing and a multi-modal data fusion technology, tumor microenvironment features and drug metabolism features are extracted, a comprehensive feature vector is generated, the effectiveness of a chemotherapy regimen is predicted by using a machine learning model, and when the effectiveness of the chemotherapy regimen is low, the chemotherapy regimen is identified. Newest clinical tests and drug data are introduced regularly, model parameters are dynamically optimized in combination with the chemotherapy regimen improvement effect, it is ensured that the model reflects the newest clinical progress in real time, the problems of model updating lag and drug innovation mismatching are effectively solved, the accuracy and effectiveness of a treatment regimen are improved, and the treatment cost is reduced. And meanwhile, side effects caused by blind attempts are reduced, and powerful support is provided for personalized precise medical treatment.

Description

Technical Field

[0001] The present invention relates to the technical field of tumor disease assessment, and particularly relates to a method for tumor disease assessment based on big data. Background Art

[0002] Tumor disease assessment based on big data refers to a method of using large-scale and multi-dimensional data, through technical means such as data mining, statistical analysis, and artificial intelligence, to conduct risk prediction, diagnosis, treatment effect evaluation, and personalized medical advice for tumor diseases. This method not only relies on traditional medical data (such as pathology, imaging, genetic testing, etc.), but also integrates big data resources such as social behavior, environmental factors, and living habits, so as to provide a more comprehensive disease management and intervention strategy. For example, in the treatment of breast cancer, through the big data analysis of integrating the patient's genomic data, imaging data (such as MRI or CT scans), treatment plans, and disease course progress, the effectiveness of a certain chemotherapy plan for the patient can be predicted. Such a big data-based model can search for efficient treatment plans from similar cases globally, provide auxiliary decision-making for doctors, reduce blind attempts, and improve the treatment effect.

[0003] The prior art has the following deficiencies: In the application of medical big data models, the mismatch between model update lag and drug innovation is a key issue. Due to the relatively fast R & D and approval speeds of drugs and treatment plans in the medical field, while big data prediction models usually rely on historical data for training, this lag may result in the model failing to reflect the latest clinical progress. For example, the efficacy of a new immunotherapy (such as anti-PD-L1 antibody) in specific breast cancer patients has been confirmed through clinical trials, but if the model is still based on the old dataset, it may not be able to identify these patients as suitable for this therapy, thus wrongly recommending traditional chemotherapy plans. In addition, such a mismatch may not only cause patients to miss more effective and personalized treatment opportunities, but also may exacerbate unnecessary treatment side effects. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for tumor disease assessment based on big data to solve the deficiencies in the background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A method for tumor disease assessment based on big data, comprising the following steps: S1: Collect the medical data of patients through multiple data interfaces, including genomic data, medical imaging data, treatment plan data, disease course progress data, and the living habit data of patients; S2: Preprocess the collected medical data and living habit data of patients, and integrate the medical data and living habit data of patients through feature extraction and multi-modal data fusion technology; S3: Extract the tumor microenvironment features in the patient's medical data and the patient's drug metabolism features in the lifestyle data, convert them into a comprehensive feature vector, and use a machine learning model to train the integrated data to predict the effectiveness of the chemotherapy regimen; S4: When the effectiveness of the chemotherapy regimen is low, regularly obtain the latest clinical trial and drug data, and dynamically update and optimize the parameters of the big data model in combination with the improvement effect of the chemotherapy regimen effectiveness; S5: Use the optimized big data model to evaluate the newly collected patient data, generate disease risk prediction, chemotherapy regimen effectiveness evaluation and personalized treatment recommendations, and at the same time provide a visualized decision-making plan.

[0006] Preferably, in S3, after analyzing the tumor microenvironment features in the extracted patient medical data, an abnormal tumor vascular density index is generated. The method for obtaining the abnormal tumor vascular density index is as follows: Load the patient's medical images, extract the region of interest (ROI), segment the tumor region and the adjacent healthy tissue region, and for each pixel in the ROI, use the threshold segmentation method to extract the vascular region: ; where is the vascular density of the ROI, is the number of vascular pixels, is the total number of pixels in the ROI; Calculate the vascular density of the tumor region and the healthy tissue region respectively: ; ; is the vascular density of the tumor region, reflecting the tumor angiogenesis situation. Vascular Pixels in Tumor ROI is the number of pixels identified as blood vessels in the tumor region, and Total Pixels in Tumor ROI is the total number of pixels within the tumor region, representing the size of the entire tumor; is the baseline vascular density of the healthy tissue, used as a comparison standard. VascularPixels in Healthy ROI is the number of pixels identified as blood vessels in the healthy tissue region, and Total Pixels inHealthy ROI is the total number of pixels in the healthy tissue region, representing the size of the healthy tissue used as the baseline control; Calculate the abnormal tumor vascular density index, that is, calculate the standardized difference in vascular density between the tumor region and the healthy region: ; is the standard deviation of the vascular density in the healthy tissue region, is the abnormal tumor vascular density index.

[0007] Preferably, in S3, after analyzing the patient's drug metabolism features in the extracted patient lifestyle data, a drug clearance rate fluctuation index is generated. The method for obtaining the drug clearance rate fluctuation index is as follows: Initial calculation of the clearance rate CL: ; D is the administered dose, AUC is the area under the drug concentration-time curve, calculated by numerical integration; Bayesian regression is used to model the relationship between the drug clearance rate CL and the individual characteristics X: ; where is the clearance rate of the i-th patient, is the j-th characteristic of the i-th patient, , are the model parameters, is the observation error; it is assumed that the regression coefficients follow a Gaussian distribution: ; where is the prior mean; is the prior variance, representing uncertainty; the prior of the noise variance: ; a, β are hyperparameters that control the shape of the prior distribution of the noise variance; According to Bayes' theorem, calculate the posterior distribution , and the expression is: ; where is the likelihood function, is the prior distribution, is the evidence. Using the parameter estimates in the posterior distribution, according to the variance of the clearance rate, define the drug clearance rate variability index DCRVI, and the expression is: ; where is the drug clearance rate variability index, is the variance of the clearance rate, is the mean of the clearance rate.

[0008] Preferably, in S3, convert the tumor vascular density abnormality index and the drug clearance rate variability index into a comprehensive feature vector, use the comprehensive feature vector as the input of the machine learning model, and use the machine learning model to predict the effectiveness value label of the chemotherapy regimen as the prediction target, and use minimizing the sum of the prediction errors of the effectiveness value labels of all chemotherapy regimens as the training target to train the machine learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the effectiveness value of the chemotherapy regimen according to the model output result, where the machine learning model is a polynomial regression model.

[0009] Preferably, compare the obtained effectiveness value of the chemotherapy regimen with the reference threshold of the effectiveness value of the chemotherapy regimen preset according to historical tumor treatment data. If the effectiveness value of the chemotherapy regimen is greater than or equal to the reference threshold of the effectiveness value of the chemotherapy regimen preset, it indicates that the effectiveness of the chemotherapy regimen is high. At this time, generate a normal signal for the chemotherapy regimen, indicating that the chemotherapy regimen is suitable for the current patient and the effect is significant. If the effectiveness value of the chemotherapy regimen is less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, it indicates that the effectiveness of the chemotherapy regimen is low. At this time, generate an abnormal signal for the chemotherapy regimen and adjust the chemotherapy regimen.

[0010] Preferably, in S4, when the effectiveness of the chemotherapy regimen is low, that is, the effectiveness value of the chemotherapy regimen generated within a fixed time period is less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, regularly obtain the latest clinical trial and drug data, collect the effectiveness values generated in the subsequent time period that are less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, and establish a corresponding data set, calculate the mean and standard deviation of the data set, and judge the improvement effect of the effectiveness of the chemotherapy regimen after analyzing it.

[0011] Preferably, if the mean of the effectiveness values in the data set is greater than or equal to the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values is less than the reference threshold of the standard deviation of the effectiveness values, generate a first-level warning signal at this time, indicating that the mean effectiveness of the chemotherapy regimen reaches the reference level and the volatility is low, indicating that the chemotherapy regimen tends to be stable and effective, but still needs to be focused on; If the mean of the effectiveness values is greater than or equal to the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values is greater than or equal to the reference threshold of the standard deviation of the effectiveness values, generate a second-level warning signal at this time, indicating that the mean effectiveness of the chemotherapy regimen reaches the reference level, but the volatility is large and there are unstable factors, and the plan details need to be adjusted; If the mean of the effectiveness values is less than the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values is greater than or equal to the reference threshold of the standard deviation of the effectiveness values, generate a third-level warning signal at this time, indicating that the mean effectiveness of the chemotherapy regimen does not reach the reference level and the volatility is large, indicating that the plan needs to be optimized or replaced; If the mean of the effectiveness values is less than the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values is less than the reference threshold of the standard deviation of the effectiveness values, no warning signal is generated at this time, indicating that the mean effectiveness of the chemotherapy regimen does not reach the reference level, but the volatility is low and the strategy needs to be replaced.

[0012] In the above technical solution, the technical effects and advantages provided by the present invention: 1. The present invention solves the problem of the lag in the update of traditional medical big data models and the mismatch with drug innovation through a big data-based tumor disease assessment method. By integrating the patient's medical data and lifestyle data through multi-modal data fusion technology, tumor microenvironment features and drug metabolism features are extracted, a comprehensive feature vector is generated, and a machine learning model is used to accurately predict the effectiveness of chemotherapy regimens. When the effectiveness of the chemotherapy regimen is low, the latest clinical trial and drug data are regularly introduced to dynamically update the model parameters to ensure that the model reflects the latest clinical progress in real time, so as to recommend the most suitable chemotherapy regimen for patients, effectively improve the treatment effect, and reduce unnecessary side effects.

[0013] 2. Through the dynamic monitoring and early warning mechanism of the effectiveness value of the chemotherapy regimen, the present invention generates early warning signals at different levels to help medical staff quickly identify and adjust unstable or inefficient chemotherapy regimens. The optimized model can not only provide disease risk prediction, chemotherapy regimen evaluation and personalized treatment recommendations, but also improve the operability of the model in clinical practice through a visual decision-making scheme, providing scientific support for the precise medical treatment of patients. This method significantly improves the accuracy, stability and personalization level of chemotherapy regimens, and promotes the intelligent and modern development of tumor treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0015] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0017] Embodiment, please refer to Figure 1 As shown, the big data-based tumor disease assessment method described in this embodiment includes the following steps: S1: Collect the patient's medical data through multiple data interfaces, including genomic data, medical image data, treatment plan data, disease course progress data, and the patient's lifestyle data; S2: Preprocess the medical data and lifestyle data of the collected patients, and integrate the medical data and lifestyle data of the patients through feature extraction and multimodal data fusion technology; S3: Extract the tumor microenvironment features in the patient's medical data and the patient's drug metabolism features in the lifestyle data, convert them into comprehensive feature vectors, and use a machine learning model to train the integrated data to predict the effectiveness of the chemotherapy regimen; S4: When the effectiveness of the chemotherapy regimen is low, regularly obtain the latest clinical trial and drug data, and dynamically update and optimize the parameters of the big data model in combination with the improvement effect of the chemotherapy regimen effectiveness; S5: Use the optimized big data model to evaluate the newly collected patient data, generate disease risk prediction, chemotherapy regimen effectiveness evaluation and personalized treatment recommendations, and provide a visual decision-making plan at the same time.

[0018] In S1, the genomic data collection includes: Whole Genome Sequencing (WGS): Provide the whole gene sequence information of the patient, covering all coding and non-coding regions. Whole Exome Sequencing (WES): Focus on the variations in the coding region sequences (such as single nucleotide polymorphisms or mutations). RNA Sequencing (RNA-seq): Provide gene expression level data for analyzing transcriptional level differences. Specific gene detection: Such as BRCA1 / BRCA2, HER2 or TP53 mutation detection. Data interfaces and sources: Gene testing laboratories: Obtain the patient's gene sequencing results through a standardized API interface (such as HL7 FHIR). Clinical databases: Extract data from the institutional databases storing the patient's gene test results. Genomic data is used to analyze the patient's genetic background, tumor driver gene mutations, and polymorphisms of drug metabolism-related genes, and to predict the effectiveness and drug resistance of the chemotherapy regimen.

[0019] Medical imaging data collection includes: MRI (Magnetic Resonance Imaging): Used for soft tissue resolution and volume measurement of tumors. CT (Computed Tomography): Provide three-dimensional structural information of tumors to evaluate size and location. PET-CT (Positron Emission Tomography): Used to analyze the metabolic activity of tumors. Ultrasound and X-ray: Assist in diagnosing tumor characteristics. Hospital PACS system (Picture Archiving and Communication System): Obtain the patient's imaging data through the DICOM protocol interface. Medical imaging sharing platforms: Such as the Open Data Set (TCIA). Medical imaging data is used for tumor localization, boundary delineation, dynamic change monitoring, and radiomics feature extraction to assist in predicting the chemotherapy response.

[0020] Treatment plan data collection includes: Chemotherapy plan: types of drugs (such as doxorubicin, paclitaxel), dosage, treatment cycles and methods (intravenous injection or oral administration). Combined treatment plan: combination of chemotherapy with radiotherapy, targeted therapy or immunotherapy. Side effect management: record adverse reactions caused by chemotherapy (such as nausea, leukopenia). Data interfaces and sources: Medical electronic health record system (EHR): obtain electronic documents of treatment plans through the FHIR interface. Clinical research data platform: obtain drug usage information of relevant trials. The treatment plan data is combined with the patient's genetic and disease course data to evaluate the applicability and efficacy of the current plan.

[0021] Disease course progression data collection includes: Changes in tumor size: obtain the reduction or increase of the tumor through imaging data and regular follow-up records. Pathological examination results: including histological classification, grading, immunohistochemical detection (such as ER, PR and HER2 expression status). Follow-up data: record the survival status, recurrence and long-term efficacy of the patient after treatment. Data interfaces and sources: Hospital LIS system (Laboratory Information System): extract pathological examination results. Follow-up management system: collect long-term monitoring data. The disease course progression data is used to analyze the efficacy of chemotherapy, the dynamic condition of the disease and the long-term prognosis assessment.

[0022] Lifestyle data collection includes: Eating habits: such as high-fat diet, frequency of alcohol intake. Exercise and weight: record the daily activity level and changes in body mass index (BMI). Smoking and drinking history: evaluate behavioral factors that may be related to tumor risk or treatment effect. Environmental exposure: such as long-term exposure to high radiation or polluted environment. Data interfaces and sources: Health management devices: such as wearable devices (smart bracelets) to collect exercise data. Health questionnaires or interview records: obtain information on daily living habits by having patients fill out questionnaires. By combining lifestyle data, the impact of non-genetic factors on the effectiveness of chemotherapy regimens can be analyzed, providing more accurate personalized treatment recommendations.

[0023] In this application, genomic, imaging, treatment plan, disease course progression and lifestyle data are collected through multiple data interfaces to form a multi-dimensional and comprehensive patient health data set. These data can not only enhance the prediction ability of machine learning models for the effectiveness of chemotherapy regimens, but also support personalized medical decision-making and long-term health management.

[0024] S2: Preprocess the collected medical data and lifestyle data of the patient, and integrate the medical data and lifestyle data of the patient through feature extraction and multi-modal data fusion technology.

[0025] Preprocess the medical data and lifestyle data of the collected patients to remove noise, duplicate values, and incomplete records in the data, and improve the data quality. Check for sequencing errors in the gene data, and delete or fill in the missing loci. For the imaging data, eliminate the blurred or low-resolution scan results. Check the completeness of the filled lifestyle data (such as whether there are null values in the alcohol consumption and exercise records). Convert the data with different units and dimensions to the same range for easy model processing. Genomic data: Normalize the mutation frequency and gene expression level to the range of 0-1. Normalize the pixel values (such as mapping the gray values from 0 to 255 to the range from 0 to 1). Lifestyle data: Normalize the exercise time on a daily / weekly basis, and normalize the alcohol consumption in standard units (such as grams per day).

[0026] Convert unstructured data (such as images and texts) into an analyzable structured form. Imaging data: The tensor form (such as the feature vector of a three-dimensional volume) extracted using a convolutional neural network (CNN). Genomic data: Convert it into a tabular form (such as a gene name-mutation status matrix). Lifestyle data: Extract key fields from the text questionnaire to generate numerical features.

[0027] Genomic data: Extract key mutated genes (such as BRCA1 / BRCA2, HER2) and their expression levels. Extract gene polymorphisms related to drug metabolism (such as the CYP450 family).

[0028] Imaging data: Extract anatomical features such as tumor size, shape, and boundary clarity from MRI / CT images. Use radiomics methods to extract texture features (such as contrast and uniformity in the gray-level co-occurrence matrix).

[0029] Disease progression data: Extract dynamic features of tumor changes over time (such as the tumor volume reduction rate). Extract the recurrence risk score (such as the Oncotype DX score).

[0030] Dietary habits: Such as fat intake (grams per day), antioxidant intake levels (amounts of vitamin E and C). Exercise level: Such as the number of daily steps, the duration of moderate-to-high-intensity exercise per week. Environmental exposure: Such as the time of exposure to a polluted environment (years). Smoking and alcohol consumption: Such as the smoking index (daily smoking amount × number of smoking years), alcohol intake (grams per day).

[0031] Directly splice medical data (such as genetic features, imaging features) and lifestyle data into a high-dimensional feature vector: X = [genomic features, imaging features, lifestyle features]; separately extract features from medical data and lifestyle data, project them into the latent space representation and then fuse them: Latent space representation of medical data: Hmedical = fmedical(Xmedical). Latent space representation of lifestyle data: Hlifestyle = flifestyle(Xlifestyle). Fusion representation: H = [Hmedical, Hlifestyle]. Use a neural network (such as a multi-layer perceptron MLP) to independently model different data types. Extract deep features, suitable for scenarios with relatively high dimensions of multi-modal data.

[0032] S3: Extract the tumor microenvironment features in the patient's medical data and the patient's drug metabolism features in the lifestyle data, convert them into a comprehensive feature vector, and use a machine learning model to train the integrated data to predict the effectiveness of the chemotherapy regimen.

[0033] The tumor microenvironment (Tumor Microenvironment, TME) refers to the complex interaction environment between tumor cells and surrounding non-tumor components (such as immune cells, blood vessels, stromal cells, cytokines, etc.), which significantly affects tumor progression and treatment response.

[0034] Key features of immune cell characteristics: T cell infiltration level: proportion of CD8+ T cells (cytotoxic T cells), reflecting the anti-tumor immune response. Proportion of M2 macrophages: related to immune suppression and chemotherapy resistance. Activity of NK cells (natural killer cells): predicting the tumor clearance ability. Data sources: Transcriptome sequencing data: Use gene expression profiles to analyze the proportion of immune cell subsets (such as the CIBERSORT tool). Immunohistochemistry (IHC) data: Analyze specific cell markers (such as CD8, CD68) in tissue sections by staining.

[0035] Key features of angiogenesis characteristics: Expression level of VEGF (vascular endothelial growth factor): related to tumor angiogenesis and the efficiency of chemotherapy drug delivery. Microvessel density (MVD): Evaluate the tumor blood vessel density by CD31 or CD34 staining. Data sources: IHC analysis: Extract the VEGF expression level and MVD index. Imaging data: Dynamic contrast-enhanced MRI (DCE-MRI) can reflect blood flow and vascular permeability.

[0036] Key features of cytokines and metabolites: IL-6, TNF-α: Levels of inflammatory factors, related to tumor proliferation and treatment resistance. Lactate concentration: A high-lactate environment can lead to immunosuppression, reflecting the metabolic characteristics of tumors. Data sources: Blood biochemical tests: Detect the concentration of specific metabolites through the patient's plasma or tissue fluid. Gene expression data: Extract the transcriptional levels of the IL6 and TNF genes.

[0037] Matrix features: Matrix rigidity: Related to the risk of tumor invasion and metastasis. Fibroblast activity (CAF): Analyze CAF markers (such as α-SMA) through gene or protein expression. Data sources: Genomic data: Extract the expression levels of key genes related to the matrix (such as COL1A1, MMP2). Imaging data: Use MRI to evaluate tumor hardness or matrix heterogeneity.

[0038] Drug metabolism characteristics reflect the patient's processes of drug absorption, distribution, metabolism, and excretion (ADME), significantly affecting the efficacy and toxicity of chemotherapeutic drugs.

[0039] Key genes of drug-metabolizing enzyme gene polymorphisms: CYP450 family: Such as CYP2D6, CYP3A4, which affect the drug metabolism rate (slow metabolizers may accumulate drug toxicity). UGT1A1: Related to the metabolism and toxicity risk of drugs such as irinotecan. Feature extraction methods: Gene detection data: Extract SNP site information of drug-metabolizing enzyme genes through whole-genome sequencing or whole-exome sequencing. Functional prediction tools: Such as PolyPhen or SIFT, to analyze the impact of gene polymorphisms on metabolic function.

[0040] Key genes of drug transporter characteristics: ABCB1 (MDR1): Encodes P-glycoprotein, related to drug efflux. SLCO1B1: Affects drug uptake in the liver. Data sources: Mutation status of transporter genes in genomic data. mRNA expression levels in transcriptome data.

[0041] Key features of liver and kidney function: Liver enzyme activity: Levels of ALT, AST (alanine aminotransferase, aspartate aminotransferase). Renal function index: Creatinine clearance rate (eGFR). Data sources: Blood test data: Obtain the liver and kidney function parameters of the patient through routine tests. Imaging data: Such as kidney ultrasound to evaluate structural integrity.

[0042] Key features of pharmacokinetic parameters: Drug elimination half-life: The rate of decline of drug concentration in plasma. Volume of drug distribution in the body: Reflects the distribution range of drugs throughout the body. Data sources: Clinical pharmacokinetic experiments: Measure the plasma drug concentration of the patient at different time points.

[0043] Key features of metabolic effects related to lifestyle habits: Dietary factors: Whether a high-fat diet induces an increase in CYP3A4 activity. Smoking behavior: Nicotine metabolism affects the activity of drug-metabolizing enzymes (such as CYP1A2). Data source: Health questionnaire data combined with blood tests.

[0044] After analyzing the tumor microenvironment features in the extracted patient medical data, an abnormal tumor vascular density index is generated. The method for obtaining the abnormal tumor vascular density index is as follows: Load the patient's medical images (such as DCE-MRI, CT, or PET-CT). Extract the region of interest (ROI): Segment the tumor region (Tumor ROI) and the adjacent healthy tissue region (Healthy ROI), which can be achieved through an automatic segmentation algorithm (such as UNet) or manual annotation. Normalize the image gray values, scaling the range to 0, 1 to reduce the influence of brightness differences: ; where is the normalized image, 、 are the minimum and maximum values of the image.

[0045] For each pixel within the ROI, use the threshold segmentation method to extract the vascular region: ; where is the vascular density of the ROI, is the number of vascular pixels (pixels with gray values exceeding a certain threshold), is the total number of pixels in the ROI.

[0046] Calculate the vascular density of the tumor region and the healthy tissue region respectively: ; ; is the vascular density of the tumor region, reflecting tumor angiogenesis. Vascular Pixels in Tumor ROI is the number of pixels identified as blood vessels in the tumor region, and Total Pixels in Tumor ROI is the total number of pixels within the tumor region, representing the size of the entire tumor; is the baseline vascular density of the healthy tissue, serving as a comparison standard. VascularPixels in Healthy ROI is the number of pixels identified as blood vessels in the healthy tissue region, and Total Pixels inHealthy ROI is the total number of pixels in the healthy tissue region, representing the size of the healthy tissue as a baseline control. Calculate the abnormal tumor vascular density index, that is, calculate the standardized difference in vascular density between the tumor region and the healthy region: ; is the standard deviation of the vascular density in the healthy tissue region, is the abnormal index of tumor vascular density.

[0047] The Tumor Vascular Density Abnormal Index (TVDAI) reflects the degree of abnormality in the vascular distribution within the tumor region. An overly large TVDAI usually indicates an extremely uneven vascular density distribution in the tumor, possibly with the presence of abnormally formed blood vessels with excessive generation. Although the number of these abnormal blood vessels is large, they often have a disordered structure and low function, resulting in the inability of drugs to be evenly distributed throughout the tumor. As a result, chemotherapy drugs are difficult to effectively penetrate into the deep regions of the tumor, thereby reducing the overall effectiveness of the chemotherapy regimen and potentially causing drug concentration in certain areas to trigger toxic reactions.

[0048] On the other hand, an overly small TVDAI may indicate insufficient angiogenesis in the tumor region, leading to limited blood flow supply. In this case, although the drug tolerance of the tumor may be weakened, the drug delivery ability will also decrease significantly, affecting the treatment effect. In addition, low vascular density may be associated with a hypoxic environment in the tumor, which activates the tumor's drug resistance mechanisms, such as enhancing the survival ability of tumor cells or inducing metastatic potential. Therefore, whether the TVDAI is too high or too low, it may indicate limited effectiveness of the chemotherapy regimen, and it is recommended to optimize the treatment strategy by combining other tumor characteristics and the individual situation of the patient.

[0049] After analyzing the patient's drug metabolism characteristics in the extracted patient lifestyle data, a drug clearance rate fluctuation index is generated. The method for obtaining the drug clearance rate fluctuation index is as follows: The input data includes: drug concentration-time data: the plasma drug concentration (Ct) of the patient at different time points. Individual characteristic data: lifestyle data (such as smoking index, alcohol consumption), liver and kidney function indicators (such as eGFR). Drug parameters: dose (D), dosing interval (T), volume of distribution (Vd). All input variables are normalized so that their ranges are scaled to [0,1]; Preliminary calculation of the clearance rate CL: ; D is the administered dose, and AUC is the area under the drug concentration-time curve, which is calculated by numerical integration.

[0050] Bayesian regression modeling of the relationship between the drug clearance rate CL and individual characteristics X: ; where is the clearance rate of the i-th patient, is the j-th characteristic of the i-th patient (such as smoking index, kidney function index). , are model parameters (regression coefficients), is the observation error.

[0051] Set the regression coefficients to follow a Gaussian distribution: ; where is the prior mean, usually taken as 0; is the prior variance, representing uncertainty. Prior of noise variance: ; a, β are hyperparameters that control the shape of the prior distribution of the noise variance.

[0052] According to Bayes' theorem, calculate the posterior distribution , and the expression is: ; in the formula, is the likelihood function, is the prior distribution, is the evidence, which is used for normalization.

[0053] Using the parameter estimation value in the posterior distribution, according to the variance of the clearance rate, define the drug clearance rate variability index DCRVI, and the expression is: ; in the formula, is the drug clearance rate variability index, is the variance of the clearance rate, is the mean of the clearance rate.

[0054] The larger the drug clearance rate variability index (DCRVI), the more significant the instability of the patient's drug clearance rate usually indicates. This volatility may lead to too high or too low drug concentrations in the body, increasing the risk of chemotherapy regimen failure. For example, when the clearance rate is too fast, the drug may not be able to maintain a sufficient effective concentration, reducing the killing effect on tumor cells; while too slow clearance rate may lead to drug accumulation in the body, increasing the probability of occurrence of side effects. A too large DCRVI often indicates that the chemotherapy regimen needs to be adjusted in dose or monitored, and even combined with other treatment methods to improve efficacy and safety.

[0055] The smaller the drug clearance rate variability index (DCRVI), the more stable the patient's drug clearance rate indicates. This usually helps to maintain the steady-state concentration of the drug in the body, ensure the full play of the therapeutic effect of the drug, and is not likely to cause dose-dependent side effects. In this case, the effectiveness of the chemotherapy regimen may be higher, and the individual's response to the standardized dose is more ideal. However, too low DCRVI may also reflect a general reduction in the patient's metabolic capacity (such as impaired liver and kidney function), and further evaluation is needed to ensure that drug metabolism and elimination are not restricted.

[0056] Convert the tumor vascular density abnormality index and the drug clearance rate fluctuation index into a comprehensive feature vector, and use the comprehensive feature vector as the input of the machine learning model. The machine learning model takes predicting the effectiveness value label of the chemotherapy regimen for each group of comprehensive feature vectors as the prediction target, and takes minimizing the sum of the prediction errors of the effectiveness value labels of all chemotherapy regimens as the training target to train the machine learning model until the sum of the prediction errors reaches convergence and then stops the model training. Determine the effectiveness value of the chemotherapy regimen according to the model output result, where the machine learning model is a polynomial regression model.

[0057] The method for obtaining the effectiveness value of the chemotherapy regimen is as follows: Obtain the corresponding function expression from the comprehensive feature vector training data of the trained machine learning model: ; In the formula, is the output function of the model, is the tumor vascular density abnormality index, is the drug clearance rate fluctuation index, is the effectiveness value of the chemotherapy regimen.

[0058] Compare the obtained effectiveness value of the chemotherapy regimen with the reference threshold of the effectiveness value of the chemotherapy regimen preset according to the historical tumor treatment data. If the effectiveness value of the chemotherapy regimen is greater than or equal to the reference threshold of the effectiveness value of the chemotherapy regimen preset, it indicates that the effectiveness of the chemotherapy regimen is high. At this time, generate a normal signal for the chemotherapy regimen, and the chemotherapy regimen is suitable for the current patient and has a significant effect. If the effectiveness value of the chemotherapy regimen is less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, it indicates that the effectiveness of the chemotherapy regimen is low. At this time, generate an abnormal signal for the chemotherapy regimen, and the chemotherapy regimen needs to be adjusted or combined with other treatment strategies.

[0059] S4: When the effectiveness of the chemotherapy regimen is low, regularly obtain the latest clinical trial and drug data, and dynamically update and optimize the parameters of the big data model in combination with the improvement effect of the effectiveness of the chemotherapy regimen.

[0060] When the effectiveness of the chemotherapy regimen is low, that is, the effectiveness value of the chemotherapy regimen generated within a fixed time period is less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, regularly obtain the latest clinical trial and drug data, collect the effectiveness values generated in the subsequent time period that are less than the reference threshold of the effectiveness value of the chemotherapy regimen preset, and establish a corresponding data set, calculate the mean and standard deviation of the data set, and judge the improvement effect of the effectiveness of the chemotherapy regimen after analyzing it.

[0061] If the mean of the validity values in the data set is greater than or equal to the reference threshold of the mean of the validity values, and the standard deviation of the validity values is less than the reference threshold of the standard deviation of the validity values, a first-level warning signal is generated at this time, indicating that the mean effectiveness of the chemotherapy regimen reaches the reference level and the volatility is low, indicating that the chemotherapy regimen tends to be stable and effective, but still needs to be focused on.

[0062] If the mean of the validity values is greater than or equal to the reference threshold of the mean of the validity values, and the standard deviation of the validity values is greater than or equal to the reference threshold of the standard deviation of the validity values, a second-level warning signal is generated at this time, indicating that: the mean effectiveness of the chemotherapy regimen reaches the reference level, but the volatility is large, and there may be unstable factors, and the details of the regimen need to be adjusted.

[0063] If the mean of the validity values is less than the reference threshold of the mean of the validity values, and the standard deviation of the validity values is greater than or equal to the reference threshold of the standard deviation of the validity values, a third-level warning signal is generated at this time, indicating that the mean effectiveness of the chemotherapy regimen does not reach the reference level and the volatility is large, indicating that the regimen needs to be optimized or replaced.

[0064] If the mean of the validity values is less than the reference threshold of the mean of the validity values, and the standard deviation of the validity values is less than the reference threshold of the standard deviation of the validity values, no warning signal is generated at this time, indicating that: the mean effectiveness of the chemotherapy regimen does not reach the reference level, but the volatility is low, and the strategy may need to be replaced, and the effect improvement is limited.

[0065] It should be noted here that the importance of the first-level warning signal is greater than that of the second-level warning signal, and the importance of the second-level warning signal is greater than that of the third-level warning signal. Relevant personnel can take corresponding handling measures according to different warning signal levels.

[0066] The first-level warning signal indicates that although the chemotherapy regimen is effective, it may need to be further consolidated. Handling measures: Maintain the existing regimen and fine-tune the model parameters in combination with the latest data.

[0067] The second-level warning signal indicates that although the effectiveness of the regimen is high, it is unstable. Handling measures: Optimize the dosage regimen or combination treatment method and update the model.

[0068] The third-level warning signal indicates that the existing regimen has poor effects and is unstable. Handling measures: Significantly adjust the chemotherapy regimen or replace the drugs and retrain the model.

[0069] No warning signal indicates that the existing regimen lacks significant improvement and the model parameters need to be thoroughly optimized in combination with the latest clinical trial data.

[0070] S5: Use the optimized big data model to evaluate the newly collected patient data, generate disease risk prediction, chemotherapy regimen effectiveness evaluation and personalized treatment recommendations, and at the same time provide a visual decision-making plan.

[0071] New patient data collection includes the following data dimensions: Medical data: genomic data (such as BRCA1 / BRCA2 mutations), medical imaging data (MRI, CT), pathological data (tumor classification). Lifestyle data: diet, exercise, smoking and drinking behaviors. Treatment data: previous treatment records (drug type, dosage).

[0072] Standardize the input data: Align the format of the new patient data with the data used by the model. Feature extraction and fusion: Extract tumor microenvironment features, drug metabolism features, etc., and generate a comprehensive feature vector. Use the optimized big data model to evaluate the new patient data and generate the following results: Disease risk prediction includes: Prediction objective: Evaluate the tumor development and recurrence risks of new patients. Evaluation process: Input the patient's comprehensive feature vector. Output a risk probability value indicating the likelihood of disease development (e.g., 0.85 represents a high risk of 85%). Visualization scheme: Bar chart: Show the risk comparison between new patients and healthy control groups. Risk radar chart: Highlight the main features affecting the risk (such as high TVDAI, low immune infiltration).

[0073] Predict the effectiveness values of different chemotherapy regimens. Use a polynomial regression model to predict the effectiveness values of chemotherapy regimens. Output the optimal regimen and its effectiveness value, and compare it with the reference threshold. Visualization scheme: Regimen effectiveness bar chart: Display the effectiveness values of different chemotherapy regimens. Dynamic heat map: Show the impact of drug dosage adjustment on the effectiveness value.

[0074] Combine the risk prediction and chemotherapy regimen evaluation results to generate personalized treatment recommendations for patients. Generation process: Sort according to the chemotherapy regimen effectiveness value and recommend the best chemotherapy regimen. Provide combined treatment recommendations (such as chemotherapy + targeted therapy or chemotherapy + immunotherapy). Visualization scheme: Treatment path diagram: Show the combination and time nodes of the recommended treatment. Effect prediction curve: Simulate the dynamic impact of treatment on tumor risk and patient health.

[0075] An interactive dashboard that comprehensively presents the evaluation results, including the following: Disease risk assessment module: The risk level (high, medium, low) is marked with colors. Compare the risk differences between healthy people and tumor patients. Chemotherapy regimen evaluation module: Display the effectiveness values of different regimens in bar charts or pie charts. Dynamically simulate the chemotherapy effect. Treatment recommendation module: Visualize the treatment path (such as a Gantt chart of the treatment plan). Show the expected efficacy and time nodes.

[0076] Explainable Artificial Intelligence (XAI) tool: Provide a transparent analysis of the model output. Use SHAP values to explain the main influencing features in disease risk prediction (such as the importance of TVDAI for chemotherapy effect). The contribution degree of personalized features in high chemotherapy regimens (such as the impact of specific gene mutations on drug response).

[0077] Building a user-friendly interface: Users can adjust input parameters (such as drug dosage, treatment time) and view the changes in prediction results in real time. Hierarchical display: Provide concise and detailed versions of the results to meet the different needs of patients and professional physicians. Collect the actual efficacy data during the patient's treatment process and dynamically update the model. Re-evaluate the chemotherapy plan based on the patient's subsequent performance and generate new suggestions.

[0078] In this embodiment, medical data (such as genomic data, medical image data, treatment plan data, disease progression data) and lifestyle data of patients are collected through multiple data interfaces. After preprocessing, information is integrated through feature extraction and multi-modal data fusion technology to extract the tumor microenvironment features and drug metabolism features of patients, which are converted into comprehensive feature vectors. Then, a machine learning model is used to train the integrated data to predict the effectiveness of the chemotherapy plan. When the effectiveness of the chemotherapy plan is low, the latest clinical trial and drug data are regularly introduced, and the parameters of the big data model are dynamically optimized in combination with the improvement effect of the chemotherapy plan. Finally, the optimized model is used to evaluate the data of new patients, generate disease risk prediction, chemotherapy plan effectiveness evaluation and personalized treatment suggestions, and display the decision-making plan in a visual form to provide support for precision medicine.

[0079] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0080] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0081] It should be understood that the term "and / or" in this text is merely a description of the relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. Additionally, the character " / " in this text generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be understood specifically by referring to the context. Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this text can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0082] As described above, this is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application.

Claims

1. A tumor disease assessment method based on big data, characterized by: The following steps are involved: S1: Collect patients’ medical data through multiple data interfaces, including genomic data, medical imaging data, treatment plan data, disease progression data, and patients’ lifestyle data; S2: Preprocess the collected medical data and lifestyle data of the patients, and integrate them through feature extraction and multimodal data fusion technology; S3: Extract tumor microenvironment features from patient medical data and drug metabolism features from patient lifestyle data, convert them into comprehensive feature vectors, and use machine learning models to train the integrated data to predict the effectiveness of chemotherapy regimens; S4: When the effectiveness of chemotherapy is low, regularly obtain the latest clinical trial and drug data, and dynamically update and optimize the parameters of the big data model based on the improvement of the effectiveness of chemotherapy; S5: Use the optimized big data model to evaluate newly collected patient data, generate disease risk predictions, chemotherapy effectiveness evaluations, and personalized treatment recommendations, while providing visual decision-making solutions.

2. The tumor disease assessment method based on big data according to claim 1, characterized in that: In S3, the tumor microenvironment characteristics in the extracted patient medical data are analyzed to generate a tumor vascular density abnormality index. The method for obtaining the tumor vascular density abnormality index is: Load the patient's medical image, extract the region of interest (ROI), segment the tumor area and the adjacent healthy tissue area, and use the threshold segmentation method to extract the vascular area for each pixel in the ROI: ; In the formula, is the blood vessel density of ROI, is the number of blood vessel pixels, is the total number of pixels in ROI; the vascular density of the tumor area and the healthy tissue area are calculated separately: ; ; is the vascular density in the tumor area, reflecting the angiogenesis of the tumor, Vascular Pixels in Tumor ROI is the number of pixels identified as blood vessels in the tumor area, and Total Pixels in Tumor ROI is the total number of pixels in the tumor area, indicating the size of the entire tumor; is the baseline vascular density of healthy tissue. As a comparison standard, VascularPixels in Healthy ROI is the number of pixels identified as blood vessels in the healthy tissue area, and TotalPixels inHealthy ROI is the total number of pixels in the healthy tissue area, indicating the size of healthy tissue as a baseline control. The tumor vascular density abnormality index is calculated, that is, the standardized difference in vascular density between the tumor area and the healthy area is calculated: ; is the standard deviation of blood vessel density in healthy tissue area, It is an index of abnormal tumor vascular density.

3. The tumor disease assessment method based on big data according to claim 2, characterized in that: In S3, the drug metabolism characteristics of the patient extracted from the patient's living habit data are analyzed to generate a drug clearance rate fluctuation index. The method for obtaining the drug clearance rate fluctuation index is: Preliminary calculation of the clearance rate CL: ; D is the dosage, AUC is the area under the drug concentration-time curve, calculated by numerical integration; Bayesian regression modeling of the relationship between drug clearance rate CL and individual characteristics X: ; In the formula, is the clearance rate of the ith patient, is the jth feature of the i-th patient, , are model parameters, is the observation error; the regression coefficient is set to conform to the Gaussian distribution: ; In the formula, is the prior mean; is the prior variance, indicating uncertainty; the noise variance prior: ; a, β are hyperparameters that control the shape of the prior distribution of the noise variance; According to Bayes' theorem, calculate the posterior distribution , the expression is: ; In the formula, is the likelihood function, is the prior distribution, As evidence, the parameter estimates in the posterior distribution are used to define the drug clearance rate fluctuation index DCRVI according to the variance of the clearance rate. The expression is: ; In the formula, is the drug clearance rate fluctuation index, is the variance of the removal rate, is the mean clearance rate.

4. The tumor disease assessment method based on big data according to claim 3 is characterized in that: In S3, the abnormal tumor vascular density index and the drug clearance rate fluctuation index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as inputs of the machine learning model. The machine learning model predicts the effectiveness value label of the chemotherapy regimen with each group of comprehensive feature vectors as the prediction target, and minimizes the sum of the prediction errors of the effectiveness value labels of all chemotherapy regimens as the training target. The machine learning model is trained until the sum of the prediction errors converges, and the model training is stopped. The effectiveness value of the chemotherapy regimen is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

5. The tumor disease assessment method based on big data according to claim 4 is characterized in that: The obtained effectiveness value of the chemotherapy regimen is compared with the reference threshold of the chemotherapy regimen effectiveness value pre-set according to the historical tumor treatment data. If the effectiveness value of the chemotherapy regimen is greater than or equal to the pre-set reference threshold of the chemotherapy regimen effectiveness value, it means that the effectiveness of the chemotherapy regimen is high. At this time, a normal chemotherapy regimen signal is generated, and the chemotherapy regimen is suitable for the current patient and has a significant effect. If the effectiveness value of the chemotherapy regimen is less than the pre-set reference threshold of the chemotherapy regimen effectiveness value, it means that the effectiveness of the chemotherapy regimen is low. At this time, an abnormal chemotherapy regimen signal is generated and the chemotherapy regimen is adjusted.

6. The tumor disease assessment method based on big data according to claim 1, characterized in that: In S4, when the effectiveness of the chemotherapy regimen is low, that is, the effectiveness value of the chemotherapy regimen generated within a fixed time period is less than a pre-set reference threshold of the effectiveness value of the chemotherapy regimen, the latest clinical trial and drug data are regularly obtained, and the effectiveness values ​​generated in subsequent time periods that are less than the pre-set reference threshold of the effectiveness value of the chemotherapy regimen are collected, and a corresponding data set is established, the mean and standard deviation of the data set are calculated, and the data set is analyzed to determine the improvement effect of the effectiveness of the chemotherapy regimen.

7. The tumor disease assessment method based on big data according to claim 6, characterized in that: If the mean of the effectiveness values ​​in the data set is greater than or equal to the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values ​​is less than the reference threshold of the standard deviation of the effectiveness values, a first-level warning signal is generated, indicating that the mean of the effectiveness of the chemotherapy regimen has reached the reference level and has low volatility, indicating that the chemotherapy regimen tends to be stable and effective, but still needs to be paid attention to; If the mean of the effectiveness values ​​is greater than or equal to the reference threshold of the mean of the effectiveness values, and the standard deviation of the effectiveness values ​​is greater than or equal to the reference threshold of the standard deviation of the effectiveness values, a secondary warning signal is generated, indicating that the mean of the effectiveness of the chemotherapy regimen has reached the reference level, but the volatility is large and there are unstable factors, and the details of the regimen need to be adjusted; If the mean of the effectiveness value is less than the reference threshold of the mean of the effectiveness value, and the standard deviation of the effectiveness value is greater than or equal to the reference threshold of the standard deviation of the effectiveness value, a third-level warning signal is generated, indicating that the mean of the effectiveness of the chemotherapy regimen has not reached the reference level and has large fluctuations, indicating that the regimen needs to be optimized or replaced; If the mean effectiveness value is less than the reference threshold of the mean effectiveness value, and the standard deviation of the effectiveness value is less than the reference threshold of the standard deviation of the effectiveness value, no warning signal is generated, indicating that the mean effectiveness of the chemotherapy regimen has not reached the reference level, but the volatility is low, and the strategy needs to be changed.

Images