An artificial intelligence-based method for detecting anti-tumor drug resistance

Abstract

The present invention relates to the field of biomedical technology and discloses an artificial intelligence-based anti-tumor drug resistance detection method and system, comprising: collecting genomic data of tumor cells and analyzing genetic changes in tumor cells; extracting tumor molecular markers from tumor cells and constructing a knowledge graph of anti-tumor drug resistance in tumor patients; extracting gene-protein features of tumor cells and identifying the pharmacochemical properties of anti-tumor drugs; fusing gene-protein features and pharmacochemical properties to perform dimensionality reduction processing to construct a drug resistance analysis model for tumor cells; analyzing drug-target interaction relationships in tumor patients and extracting gene-resistance association features of tumor cells to identify the model reliability of the drug resistance analysis model; defining the result interpretation of the drug resistance analysis model, and combining the model reliability and result interpretation to output the anti-tumor drug resistance detection results of tumor patients. The present invention can improve the accuracy of anti-tumor drug resistance detection.

Description

Technical Field

[0001] The present invention relates to an anti-tumor drug resistance detection method and system based on artificial intelligence, belonging to the field of biomedical technology. Background Art

[0002] Antitumor drug resistance refers to the phenomenon that tumor cells are insensitive to antitumor drugs. During the antitumor treatment process, once tumor cells develop drug resistance, the therapeutic effect of antitumor drugs will be significantly reduced, leading to the failure of antitumor treatment. With the continuous development and iteration of antitumor drugs, the drug resistance problem has become more and more prominent. Therefore, detecting the drug resistance of tumor cells and in-depth understanding of their resistance mechanism are of great significance for guiding clinical treatment and formulating personalized treatment plans.

[0003] Traditional anti-tumor drug resistance detection mainly uses in vitro drug sensitivity testing methods. By co-culturing tumor cells with different concentrations of anti-tumor drugs and observing cell growth, the sensitivity of the drug is determined. However, the experimental results under this method may differ from the in vivo situation, such as cell status, making it difficult to fully predict the patient's drug resistance, thereby reducing the accuracy of anti-tumor drug resistance detection.

[0004] Therefore, a solution is urgently needed to improve the accuracy of anti-tumor drug resistance detection. Summary of the Invention

[0005] The present invention provides an artificial intelligence-based anti-tumor drug resistance detection method and system, the main purpose of which is to improve the accuracy of anti-tumor drug resistance detection.

[0006] To achieve the above objectives, the present invention provides an artificial intelligence-based anti-tumor drug resistance detection method, comprising:

[0007] Obtaining an anti-tumor drug to be tested and its corresponding tumor patient, identifying tumor cells of the tumor patient, extracting tumor molecular markers of the tumor cells, collecting genomic data of the tumor cells, and analyzing genetic changes in the tumor cells based on the genomic data;

[0008] Collecting drug structure data and drug target data of the anti-tumor drug, and collecting drug sensitivity data of the tumor cells, and combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers to construct a knowledge graph of anti-tumor drug resistance of the tumor patient;

[0009] extracting gene-protein characteristics of the tumor cells based on the gene changes, identifying the pharmacochemical properties of the anti-tumor drug based on the drug structure data, and calculating the degree of correlation between the gene-protein characteristics and the pharmacochemical properties;

[0010] According to the degree of association, the gene-protein feature and the drug chemical property are subjected to fusion dimensionality reduction processing to obtain a fusion feature, and based on the fusion feature, a drug resistance analysis model of the tumor cell is constructed;

[0011] Based on the anti-tumor drug resistance knowledge graph, the drug-target interaction relationship of the tumor patient is analyzed, and the gene-drug resistance association characteristics of the tumor cells are extracted. According to the drug-target interaction relationship and the gene-drug resistance association characteristics, the model reliability of the drug resistance analysis model is identified;

[0012] According to the anti-tumor drug resistance knowledge graph, the result interpretation of the drug resistance analysis model is defined, and the anti-tumor drug resistance detection result of the tumor patient is output in combination with the reliability of the model and the result interpretation.

[0013] Optionally, analyzing the genetic changes of the tumor cells based on the genomic data includes:

[0014] extracting tumor suppressor genes and viral oncogenes of the tumor cells based on the genomic data;

[0015] Detecting the functional impairment of the tumor suppressor gene and detecting the abnormal activation state of the viral oncogene;

[0016] Analyzing changes in expression levels of the tumor suppressor gene and the viral oncogene based on the functional impairment manifestations and the abnormal activation state;

[0017] Identifying factors influencing the changes in expression levels and collecting molecular genetic data corresponding to the influencing factors;

[0018] The gene changes of the tumor cells are analyzed by combining the expression level changes and the molecular genetics data.

[0019] Optionally, the combining of the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker to construct the anti-tumor drug resistance knowledge graph of the tumor patient includes:

[0020] Determine the disease type of the tumor patient and its corresponding field range by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker;

[0021] Based on the disease type, extracting the drug structure data, the drug target data, the drug sensitivity data and the core keywords of the tumor molecular markers;

[0022] Extracting entity categories corresponding to the disease types based on the core keywords;

[0023] Based on the domain scope, collecting domain knowledge of the disease type;

[0024] Identifying the relationship patterns between the entity categories based on the domain knowledge and extracting attribute features of the entity categories;

[0025] Based on the entity categories, the relationship patterns and the attribute characteristics, a knowledge graph of the anti-tumor drug resistance of the tumor patients is constructed.

[0026] Optionally, extracting gene-protein features of the tumor cells based on the gene changes includes:

[0027] Based on the gene changes, identifying the mutated genes of the tumor cells;

[0028] Extracting the encoded protein corresponding to the mutated gene and analyzing the status of the encoded protein;

[0029] Determining, based on the status manifestations, a gene-protein functional association relationship between the mutated gene and the protein encoding the protein;

[0030] Based on the gene-protein functional association relationship, constructing a gene-protein relationship network of the mutated gene and the protein encoding the protein;

[0031] The gene-protein characteristics of the tumor cells are extracted based on the gene-protein relationship network.

[0032] Optionally, the gene-protein feature and the drug chemical property are subjected to fusion dimensionality reduction processing according to the degree of association to obtain a fusion feature, including:

[0033] Collecting correlation data between the gene-protein feature and the drug chemical property according to the correlation degree, and extracting correlation features between the gene-protein feature and the drug chemical property;

[0034] Performing data integration processing on the associated data to obtain integrated data;

[0035] Based on the association features, identifying the disease type of the gene-protein feature, and analyzing the drug response of the disease type;

[0036] extracting key features of the gene-protein signature and the drug chemical properties from the integrated data according to the disease type and the drug response;

[0037] Execute the key feature dimensionality reduction processing to obtain the reduced dimensionality feature;

[0038] Based on the integrated data and the dimensionality reduction features, a fusion feature of the gene-protein feature and the drug chemical property is obtained.

[0039] Optionally, constructing a drug resistance analysis model for the tumor cells based on the fusion feature includes:

[0040] Identifying a tumor patient corresponding to the tumor cell, and analyzing the disease type of the tumor patient based on the fusion signature;

[0041] Extracting biomarker characteristics of the tumor cells according to the disease type;

[0042] Based on the biomarker characteristics, identifying factors affecting drug resistance of the tumor cells;

[0043] Analyzing the potential drug resistance mechanism of the tumor cells based on the factors affecting drug resistance;

[0044] Combining the disease type, the drug resistance influencing factors and the potential drug resistance mechanism, a drug resistance analysis model for the tumor cells is constructed.

[0045] Optionally, analyzing the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph includes:

[0046] Based on the anti-tumor drug resistance knowledge graph, identifying the anti-cancer drugs for the tumor patient and their corresponding target proteins;

[0047] extracting structural features of the anticancer drug and the target protein;

[0048] Identifying the binding site between the anticancer drug and the target protein based on the structural characteristics;

[0049] Based on the binding site, performing molecular docking processing between the anticancer drug and the target protein to obtain a molecular docking result;

[0050] Analyzing the binding mode between the anticancer drug and the target protein according to the molecular docking results;

[0051] Based on the binding mode and the molecular docking results, identifying the protein function of the target protein;

[0052] Combining the protein function and the binding mode, the drug-target interaction relationship of the tumor patient is analyzed.

[0053] Optionally, extracting gene-drug resistance association features of the tumor cells based on the anti-tumor drug resistance knowledge graph includes:

[0054] Extracting drug resistance genes and sensitivity genes of the tumor cells based on the anti-tumor drug resistance knowledge graph;

[0055] Analyzing the gene expression levels of the drug-resistant gene and the drug-sensitive gene;

[0056] Calculating the difference coefficient between the drug-resistant gene and the drug-sensitive gene according to the gene expression levels;

[0057] extracting the gene characteristics of the drug resistance gene based on the difference coefficient;

[0058] analyzing the drug resistance formation mechanism of the tumor cells based on the gene characteristics;

[0059] Identifying the causes of the resistance development mechanism;

[0060] Based on the causes, analyzing the drug resistance change patterns of the tumor cells;

[0061] Combining the drug resistance formation mechanism and the drug resistance change pattern, the gene-drug resistance association characteristics of the tumor cells are extracted.

[0062] Optionally, identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association characteristics includes:

[0063] Identifying the sensitivity genes corresponding to the drug-target interaction relationship, and identifying the drug resistance genes corresponding to the gene-drug resistance association characteristics;

[0064] Collecting association data of the sensitive genes and the drug-resistant genes, and classifying and labeling the association data according to the drug-target interaction relationship and the gene-drug resistance association characteristics to obtain classification label data;

[0065] Based on the classification mark data, data verification is performed on the drug resistance analysis model to obtain a data verification result;

[0066] Defining evaluation indicators of the drug resistance analysis model according to the data verification results;

[0067] Calculating the accuracy of the drug resistance analysis model based on the evaluation index;

[0068] According to the accuracy rate, the model reliability of the drug resistance analysis model is identified.

[0069] In order to solve the above problems, the present invention also provides an anti-tumor drug resistance detection system based on artificial intelligence, the system comprising:

[0070] A gene analysis module is used to obtain the anti-tumor drug to be tested and its corresponding tumor patient, identify the tumor cells of the tumor patient, extract tumor molecular markers of the tumor cells, collect genomic data of the tumor cells, and analyze the genetic changes of the tumor cells based on the genomic data;

[0071] The knowledge graph construction module is used to collect the drug structure data and drug target data of the anti-tumor drug, and collect the drug sensitivity data of the tumor cells, and combine the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers to construct the anti-tumor resistance knowledge graph of the tumor patient.

[0072] a correlation feature extraction module, configured to extract the gene-protein features of the tumor cells based on the gene changes, identify the pharmacochemical properties of the anti-tumor drug based on the drug structure data, and calculate the degree of correlation between the gene-protein features and the pharmacochemical properties;

[0073] a model generation module, configured to perform dimensionality reduction fusion processing on the gene-protein features and the drug chemical properties according to the degree of association to obtain a fusion feature, and construct a drug resistance analysis model for the tumor cells based on the fusion feature;

[0074] a model validation module for analyzing the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph, extracting the gene-drug resistance association characteristics of the tumor cells, and identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association characteristics;

[0075] The result output module is used to define the result interpretation of the resistance analysis model according to the anti-tumor resistance knowledge graph, and output the anti-tumor resistance detection result of the tumor patient in combination with the reliability of the model and the result interpretation.

[0076] Compared with the problems described in the background technology, the embodiment of the present invention can extract tumor molecular markers of the tumor cells by identifying the tumor cells of the tumor patient, collecting the genomic data of the tumor cells, and analyzing the genetic changes of the tumor cells, so as to help understand the molecular mechanism of tumor cell resistance and the heterogeneity within the tumor, and avoid the use of drugs that may induce drug resistance; further, the embodiment of the present invention can enhance the richness and accuracy of the anti-tumor drug resistance knowledge graph by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers, so as to better reveal the molecular mechanism of tumor resistance; in addition, the embodiment of the present invention can extract the gene-protein characteristics of the tumor cells based on the genetic changes, which can be used as input data for predicting tumor resistance models, monitor tumor progression and the development of drug resistance, and predict the effects of drugs with different chemical properties on specific tumor cell lines by identifying the drug chemical properties of the anti-tumor drugs; secondly, the embodiment of the present invention can predict the effects of drugs with different chemical properties on specific tumor cell lines by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers. The fusion features are used to construct a drug resistance analysis model for the tumor cells, which can dynamically monitor changes in tumor cell resistance and predict tumor cell resistance to specific anti-tumor drugs, thereby helping to screen and develop new anti-tumor drugs. Thirdly, the embodiments of the present invention analyze the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph and extract the gene-drug resistance association features of the tumor cells. This can predict the patient's possible response and efficacy to specific drugs, reveal the potential mechanism of tumor cell resistance, and identify the key genes that cause tumor cells to develop drug resistance, so as to verify the accuracy of the model, avoid incorrect decisions of the model, and improve the credibility of the model. Finally, the embodiments of the present invention output the anti-tumor drug resistance test results of the tumor patient by combining the reliability of the model and the interpretation of the results. This can guide doctors to select more suitable drugs for patients, thereby improving treatment effects, reducing the physical and economic burden of ineffective treatment, and improving the quality of life of patients. It can also help relevant personnel design new drugs that can overcome drug resistance, thereby promoting the advancement of anti-tumor drugs. Therefore, the artificial intelligence-based anti-tumor drug resistance detection method and system provided in the embodiments of the present invention can improve the accuracy of anti-tumor drug resistance detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 A schematic diagram of a process for detecting anti-tumor drug resistance based on artificial intelligence according to an embodiment of the present invention;

[0078] Figure 2 A schematic diagram of a module for implementing the artificial intelligence-based anti-tumor drug resistance detection method provided in one embodiment of the present invention.

[0079] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0080] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0081] The embodiments of the present application provide an artificial intelligence-based anti-tumor drug resistance detection method. The execution subject of the artificial intelligence-based anti-tumor drug resistance detection method includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiments of the present application. In other words, the artificial intelligence-based anti-tumor drug resistance detection method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster.

[0082] Example 1:

[0083] Reference Figure 1 FIG2 is a flow chart of an artificial intelligence-based anti-tumor drug resistance detection method according to an embodiment of the present invention. In this embodiment, the artificial intelligence-based anti-tumor drug resistance detection method includes:

[0084] S1. Obtain the anti-tumor drug to be tested and its corresponding tumor patient, identify the tumor cells of the tumor patient, and collect the genomic data of the tumor cells. Based on the genomic data, analyze the genetic changes of the tumor cells and extract tumor molecular markers of the tumor cells.

[0085] The embodiment of the present invention can provide data support for subsequent anti-tumor resistance detection by obtaining the anti-tumor drug to be tested and its corresponding tumor patient, wherein the anti-tumor drug refers to a drug used to treat tumor diseases, and the tumor patient refers to a person suffering from tumor diseases.

[0086] Furthermore, by identifying tumor cells from the tumor patient and collecting genomic data from the tumor cells, embodiments of the present invention can help understand the molecular mechanisms of tumor cell drug resistance and heterogeneity within the tumor, thereby avoiding the use of drugs that may induce drug resistance. The tumor cells refer to abnormal cells that make up the tumor, and the genomic data refers to all genetic information related to the tumor cell genome, including copy number variation, gene expression profile, etc. Optionally, tumor cell identification in the tumor patient can be achieved through a needle biopsy, and the genomic data of the tumor cells can be collected through Sanger sequencing technology.

[0087] The embodiments of the present invention analyze the genetic changes of the tumor cells based on the genomic data and extract tumor molecular markers of the tumor cells, which can be used as biomarkers for specific types of tumors to help doctors make accurate diagnoses and classifications, thereby better predicting patients' responses to specific drugs. The genetic changes refer to various genetic variations and expression changes that occur in the tumor cell genome, such as gene mutations and copy number variations. The tumor molecular markers refer to molecules that play an important role in the occurrence, development, metastasis and drug resistance of tumors, such as proteins and genes.

[0088] Optionally, the tumor molecular markers of the tumor cells extracted based on the genomic data can be obtained through pathological sections of the tumor cells.

[0089] As an embodiment of the present invention, the analyzing the genetic changes of the tumor cells based on the genomic data includes: extracting tumor suppressor genes and viral oncogenes of the tumor cells based on the genomic data; detecting functional impairment of the tumor suppressor genes, and detecting abnormal activation states of the viral oncogenes; analyzing expression level changes of the tumor suppressor genes and the viral oncogenes based on the functional impairment and the abnormal activation states; identifying influencing factors of the expression level changes, and collecting molecular genetic data corresponding to the influencing factors; and analyzing the genetic changes of the tumor cells in combination with the expression level changes and the molecular genetic data.

[0090] Among them, the tumor suppressor gene refers to a gene that inhibits tumor formation, controls the cell cycle, and promotes cell differentiation in normal cells, such as an anti-cancer gene. The viral oncogene refers to an oncogene that causes malignant transformation of cells and tumorigenesis, such as a proto-oncogene. The functional impairment manifestation refers to the phenomenon in which the function of the protein encoded by the tumor suppressor gene is weakened or lost due to mutation, deletion, methylation, or other mechanisms. The abnormal activation state refers to the state in which the activity of the protein encoded by the viral oncogene is enhanced or continuously activated due to mutation, copy number increase, overexpression, or other mechanisms. The expression level change refers to the change in the transcriptional activity of the tumor suppressor gene and the viral oncogene compared with normal cells. The influencing factors refer to various biological and environmental factors that can affect gene expression and function, including gene mutation, copy number variation, epigenetic modification (such as DNA methylation and histone modification), etc. The molecular genetic data refers to information on the structure, function, and variation of genes and genomes obtained through molecular biology techniques.

[0091] Optionally, the functional impairment of the tumor suppressor gene and the abnormal activation state of the viral oncogene can be detected through cell function experiments, and the identification of influencing factors of the expression level changes can be achieved using high-throughput sequencing technology, such as RNA sequencing technology, and the molecular genetic data corresponding to the influencing factors can be collected through a biological sample library.

[0092] S2. Collect the drug structure data and drug target data of the anti-tumor drug, and collect the drug sensitivity data of the tumor cells, and combine the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers to construct an anti-tumor drug resistance knowledge graph for the tumor patient.

[0093] The embodiments of the present invention can help identify the interaction between the drug and specific molecules in the tumor cells by collecting the drug structure data and drug target data of the anti-tumor drug, and collecting the drug sensitivity data of the tumor cells, so as to better understand the drug's mechanism of action and drug resistance. The drug structure data refers to the chemical structure information data of the anti-tumor drug, including the atomic composition of the molecule, the chemical bond connection method, etc. The drug target data refers to the object of action of the anti-tumor drug in the cell, including the identity of the target, the interaction between the drug and the target, etc. The drug sensitivity data refers to the analysis data of the response of tumor cells to specific anti-tumor drugs, including the minimum inhibitory concentration and maximum inhibitory concentration of the drug, etc.

[0094] Optionally, the drug structure data and drug target data of the anti-tumor drug can be collected using a public database, such as a drug database, and the drug sensitivity data of the tumor cells can be collected through clinical trial data.

[0095] Furthermore, the embodiments of the present invention construct an anti-tumor drug resistance knowledge graph for the tumor patient by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers, thereby enhancing the richness and accuracy of the anti-tumor drug resistance knowledge graph to better reveal the molecular mechanism of tumor resistance. The anti-tumor drug resistance knowledge graph refers to a resource library that can integrate various data and knowledge related to tumor resistance.

[0096] As an embodiment of the present invention, the combination of the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker to construct the anti-tumor drug resistance knowledge graph of the tumor patient includes: combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker to determine the disease type of the tumor patient and its corresponding field scope; based on the disease type, extracting the core keywords of the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker; according to the core keywords, extracting the entity category corresponding to the disease type; based on the field scope, collecting the field knowledge of the disease type; according to the field knowledge, identifying the relationship pattern between the entity categories and extracting the attribute characteristics of the entity categories; based on the entity categories, the relationship pattern and the attribute characteristics, constructing the anti-tumor drug resistance knowledge graph of the tumor patient.

[0097] Among them, the disease type refers to the patient's specific disease or pathological condition, such as breast cancer, brain cancer, etc., the field scope refers to the medical field to which the disease type belongs, the core keywords refer to the most critical words or terms for understanding and analyzing the disease type, including drug names, target proteins, etc., the entity category refers to the specific object corresponding to the disease type, such as drugs, genes, proteins, diseases, etc., the domain knowledge refers to the knowledge and information verified in a specific field, including scientific theories, experimental results, clinical guidelines, etc., the relationship pattern refers to the specific relationship type between entities, such as drug-target interaction, and the attribute characteristics refer to the specific attributes or characteristics of an entity, such as the chemical structure of the drug, the biological function of the target, the mutation status of the gene, etc.

[0098] Optionally, the core keyword extraction of the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker based on the disease type can be obtained through natural language processing tools, the relationship pattern recognition between the entity categories based on the domain knowledge can be implemented using a relationship extraction algorithm, the attribute feature extraction of the entity categories based on the domain knowledge can be obtained through database query, such as a gene database, and the construction of the anti-tumor drug resistance knowledge graph of the tumor patient based on the entity categories, the relationship patterns and the attribute features can be implemented using graph database technology, such as Neo4j graph database technology.

[0099] S3. Based on the gene changes, extract the gene-protein characteristics of the tumor cells, identify the drug chemical properties of the anti-tumor drug according to the drug structure data, and calculate the degree of correlation between the gene-protein characteristics and the drug chemical properties.

[0100] The embodiments of the present invention extract the gene-protein characteristics of the tumor cells based on the gene changes, which can be used as input data for predicting tumor resistance models to monitor tumor progression and the development of drug resistance. The gene-protein characteristics refer to the attribute characteristics of tumor cells in gene expression and protein expression, such as gene expression patterns, protein functional domains, etc.

[0101] As an embodiment of the present invention, the extracting of gene-protein characteristics of the tumor cells based on the gene changes includes: identifying the mutated genes of the tumor cells based on the gene changes; extracting the encoded proteins corresponding to the mutated genes, and analyzing the status performance of the encoded proteins; determining the gene-protein functional association relationship between the mutated genes and the encoded proteins based on the status performance; constructing a gene-protein relationship network between the mutated genes and the encoded proteins based on the gene-protein functional association relationship; and extracting the gene-protein characteristics of the tumor cells based on the gene-protein relationship network.

[0102] Among them, the mutant gene refers to a gene that has mutated or varied, the encoded protein refers to the protein transcribed and translated from a specific gene, the state expression refers to the various properties and behaviors of the protein in the cell, including the expression level of the protein, post-translational modification status, etc., the gene-protein functional association relationship refers to the connection between gene changes and the function of the protein it encodes, such as the mutation of a gene may cause the structure or function of the protein it encodes to change, and the gene-protein relationship network refers to a network that displays the complex interactions and regulatory relationships between genes and proteins.

[0103] Optionally, based on the gene changes, the identification of the mutated genes of the tumor cells can be obtained through gene chips, and the extraction of the coding protein corresponding to the mutated gene can be achieved using messenger RNA, such as transcribing the DNA sequence of the mutated gene into messenger RNA, which is recognized by the ribosome after processing and synthesized into the corresponding protein through the translation process. The status performance analysis of the coded protein can be obtained by analyzing the subcellular localization of the coded protein. Based on the gene-protein functional association relationship, the gene-protein relationship network construction of the mutated gene and the coding protein can be achieved using network analysis tools, such as the Cytoscape tool.

[0104] Furthermore, the embodiments of the present invention can predict the effects of drugs with different chemical properties on specific tumor cell lines by identifying the drug chemical properties of the anti-tumor drug based on the drug structure data. The drug chemical properties refer to the characteristics of the anti-tumor drug in terms of chemical structure, properties and mechanism of action, such as analytical structure, solubility, metabolic stability, etc.

[0105] Optionally, the identification of the medicinal chemical properties of the anti-tumor drug based on the drug structure data can be determined through experiments, such as mass spectrometry experiments, pKa determination, etc.

[0106] The embodiments of the present invention can reveal the molecular mechanism of drug action and guide drug design by calculating the degree of correlation between the gene-protein characteristics and the drug chemical properties. The degree of correlation refers to the strength and direction of the relationship between the quantified gene-protein characteristics and the drug chemical properties.

[0107] In an optional embodiment of the present invention, the degree of association between the gene-protein feature and the drug chemical properties is calculated using the following formula:

[0108]

[0109] Among them, P represents the degree of association between gene-protein features and drug chemical properties, N represents the total number of features corresponding to gene-protein features and drug chemical properties, a represents the feature pair index of gene-protein features and drug chemical properties, and B a ' represents the normalized value of the gene-protein feature of the a-th pair of gene-protein feature and drug chemical property feature, represents the average value of gene-protein characteristics, E a ' represents the normalized value of the medicinal chemical property corresponding to the normalized value of the gene-protein feature, Represents the average value of medicinal chemistry properties corresponding to the average value of gene-protein features.

[0110] S4. According to the degree of association, the gene-protein characteristics and the drug chemical properties are fused and reduced to obtain a fusion feature, and based on the fusion feature, a drug resistance analysis model of the tumor cells is constructed.

[0111] The embodiment of the present invention obtains fusion features by fusing and reducing the gene-protein features and the drug chemical properties according to the degree of association, thereby improving the analysis efficiency of the anti-tumor resistance analysis model, reducing the risk of model overfitting, and improving its generalization ability. The fusion features refer to combining features from different data sources or different types to form a new feature set.

[0112] As an embodiment of the present invention, the gene-protein feature and the drug chemical property are fused and reduced in dimension according to the degree of association to obtain a fusion feature, including: collecting the association data of the gene-protein feature and the drug chemical property according to the degree of association, and extracting the association feature of the gene-protein feature and the drug chemical property; performing data integration processing on the association data to obtain integrated data; based on the association feature, identifying the disease type of the gene-protein feature, and analyzing the drug response of the disease type; based on the disease type and the drug response, extracting the key features of the gene-protein feature and the drug chemical property from the integrated data; performing the key feature dimensionality reduction processing to obtain a reduced dimension feature; and obtaining the fusion feature of the gene-protein feature and the drug chemical property based on the integrated data and the reduced dimension feature.

[0113] Among them, the associated data refers to the data on the interaction and relationship between gene-protein characteristics and drug chemical properties, the associated features refer to the key attributes or indicators extracted from the associated data that can reflect the relationship between gene-protein characteristics and drug chemical properties, the integrated data refers to the comprehensive data set obtained after data integration processing, the disease type refers to the specific disease or pathological state identified based on gene-protein characteristics and drug response data, such as leukemia, the drug response refers to the patient or experimental model's treatment response to a specific drug, including efficacy, side effects and drug resistance, the key features refer to the gene-protein features and drug chemical properties extracted from the integrated data that are most important for understanding the disease type and predicting drug response, the feature dimensionality reduction processing refers to the process of obtaining a set of important features by selecting representative features and reducing the number of features under certain limited conditions, and the reduced dimensionality features refer to the features obtained through feature dimensionality reduction processing, which retain key information while reducing the data dimension.

[0114] Optionally, the data integration processing of the associated data can be implemented using Python tools, the identification of the disease type based on the associated features and the gene-protein features can be determined by gene expression spectrum analysis, the drug response analysis of the disease type based on the associated features can be implemented using randomized controlled trials, and the key feature dimensionality reduction processing can be obtained by principal component analysis.

[0115] Furthermore, the embodiment of the present invention constructs a drug resistance analysis model of the tumor cells based on the fusion characteristics, which can dynamically monitor changes in tumor cell resistance and predict the resistance of tumor cells to specific anti-tumor drugs, so as to help screen and develop new anti-tumor drugs. The drug resistance analysis model refers to a mathematical model for predicting and analyzing the resistance of tumor cells to specific anti-tumor drugs.

[0116] As an embodiment of the present invention, the drug resistance analysis model of the tumor cells is constructed based on the fusion characteristics, including: identifying the tumor patient corresponding to the tumor cells, and analyzing the disease type of the tumor patient based on the fusion characteristics; extracting the biomarker characteristics of the tumor cells according to the disease type; identifying the drug resistance influencing factors of the tumor cells based on the biomarker characteristics; analyzing the potential drug resistance mechanism of the tumor cells according to the drug resistance influencing factors; and constructing the drug resistance analysis model of the tumor cells in combination with the disease type, the drug resistance influencing factors and the potential drug resistance mechanism.

[0117] Among them, the disease type refers to the specific disease type or classification suffered by the tumor patient, such as lung cancer, breast cancer, colorectal cancer, etc., the biomarker characteristics refer to measurable properties or characteristics used to indicate specific changes in tumor cells, such as blood chemical indicators, hormone levels or metabolite concentrations, etc., the drug resistance influencing factors refer to various factors that may affect the sensitivity of tumor cells to therapeutic drugs, including differences in the activity of drug-metabolizing enzymes, changes in the tumor microenvironment, etc., and the potential drug resistance mechanism refers to the potential biological mechanism by which tumor cells develop drug resistance, such as the activation of cell signaling pathways leading to reduced resistance to drug effects.

[0118] Optionally, according to the disease type, the biomarker feature extraction of the tumor cells can be obtained by analyzing the metabolite changes of the tumor cells, and based on the biomarker features, the identification of the factors affecting the drug resistance of the tumor cells can be achieved by drug sensitivity testing. According to the factors affecting the drug resistance, the analysis of the potential drug resistance mechanism of the tumor cells can be obtained by gene editing tools, such as Cas9 gene editing tools. In combination with the disease type, the factors affecting the drug resistance and the potential drug resistance mechanism, the construction of the drug resistance analysis model of the tumor cells can be achieved by using a support vector machine.

[0119] S5. Based on the anti-tumor drug resistance knowledge graph, the drug-target interaction relationship of the tumor patient is analyzed, and the gene-resistance association characteristics of the tumor cells are extracted. According to the drug-target interaction relationship and the gene-resistance association characteristics, the model reliability of the drug resistance analysis model is identified.

[0120] The embodiments of the present invention analyze the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph, thereby predicting the patient's possible response and efficacy to a specific drug and revealing the potential mechanism by which tumor cells develop drug resistance. The drug-target interaction relationship refers to the interaction between drug molecules and specific target proteins in the organism.

[0121] As an embodiment of the present invention, the analysis of the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph includes: identifying the anticancer drugs of the tumor patient and their corresponding target proteins based on the anti-tumor drug resistance knowledge graph; extracting the structural characteristics of the anticancer drugs and the target proteins; identifying the binding sites of the anticancer drugs and the target proteins based on the structural characteristics; performing molecular docking processing of the anticancer drugs and the target proteins based on the binding sites to obtain molecular docking results; analyzing the binding mode of the anticancer drugs and the target proteins based on the molecular docking results; identifying the protein function of the target protein based on the binding mode and the molecular docking results; and analyzing the drug-target interaction relationship of the tumor patient in combination with the protein function and the binding mode.

[0122] Among them, the anti-cancer drug refers to a drug used to treat cancer, including chemotherapy drugs, targeted therapy drugs, immunotherapy drugs, etc., the target protein refers to a specific protein on which the drug acts, such as an enzyme, the structural characteristics refer to the physical and chemical properties of the anti-cancer drug and the target protein, including their three-dimensional structure, chemical composition, charge distribution, etc., the binding site refers to a specific area or position on the target protein to which the drug can bind, the molecular docking processing refers to a process of calculating the binding mode between the drug and the protein, the molecular docking result refers to the output result of the molecular docking processing, the binding mode refers to the spatial arrangement and interaction mode when the drug molecule binds to the target protein, and the protein function refers to the role of the target protein in the biological process, such as the gene expression regulation function of the target protein.

[0123] Optionally, the structural feature extraction of the anticancer drug and the target protein can be determined by nuclear magnetic resonance, and the molecular docking processing of the anticancer drug and the target protein based on the binding site can be achieved using Glid docking software, and the protein function identification of the target protein based on the binding mode and the molecular docking results can be obtained through gene knockout and knock-in experiments.

[0124] Furthermore, the embodiments of the present invention extract the gene-resistance association characteristics of the tumor cells based on the anti-tumor drug resistance knowledge graph, which helps to gain a deeper understanding of the mechanism of tumor resistance and identify the key genes that cause tumor cells to develop drug resistance. The gene-resistance association characteristics refer to characteristics related to the resistance of tumor cells to specific anti-tumor drugs, such as gene mutations.

[0125] As an embodiment of the present invention, the extracting gene-drug resistance association characteristics of the tumor cells based on the anti-tumor drug resistance knowledge graph includes: extracting the drug resistance genes and sensitivity genes of the tumor cells based on the anti-tumor drug resistance knowledge graph; analyzing the gene expression levels of the drug resistance genes and the sensitivity genes; calculating the difference coefficient between the drug resistance genes and the sensitivity genes based on the gene expression levels; extracting the gene characteristics of the drug resistance genes based on the difference coefficient; analyzing the drug resistance formation mechanism of the tumor cells based on the gene characteristics; identifying the cause of the drug resistance formation mechanism; analyzing the drug resistance change pattern of the tumor cells based on the cause; and extracting the gene-drug resistance association characteristics of the tumor cells in combination with the drug resistance formation mechanism and the drug resistance change pattern.

[0126] Among them, the drug-resistance gene refers to a gene that makes tumor cells resistant to originally effective drugs after gene mutation or abnormal expression, the sensitive gene refers to a gene that makes tumor cells more sensitive to drugs, the gene expression level refers to the amount of gene transcription into protein or RNA molecules, the variance coefficient refers to the degree of difference between the expression levels of drug-resistance genes and sensitive genes, the gene characteristics refer to the specific attributes of drug-resistance genes, such as expression pattern, mutation status, copy number changes, etc., the drug resistance formation mechanism refers to the specific biological process that leads to tumor cell resistance, including gene mutation, signal pathway changes, drug metabolism changes, etc., the cause of generation refers to the root cause behind the drug resistance formation mechanism, such as microenvironmental factors, and the drug resistance change pattern refers to the expression change pattern of drug-resistance genes under different drug pressures.

[0127] Optionally, based on the anti-tumor drug resistance knowledge graph, the extraction of drug resistance genes and sensitivity genes of the tumor cells can be achieved using data mining technology, and the gene expression levels of the drug resistance genes and the sensitivity genes can be obtained by hybridizing the probes on the microarray chip with the mRNA in the tumor cell sample. Based on the generation cause, the analysis of the change pattern of the drug resistance of the tumor cells can be achieved using time series experiments.

[0128] In an optional embodiment of the present invention, the difference coefficient between the drug-resistant gene and the sensitive gene is calculated according to the gene expression level using the following formula:

[0129]

[0130] Among them, r represents the coefficient of difference between drug-resistant genes and sensitive genes, X1 represents the gene expression level corresponding to the drug-resistant gene, X2 represents the gene expression level corresponding to the sensitive gene, S represents the combined standard deviation of the drug-resistant gene and the sensitive gene, and m represents the number of samples corresponding to the drug-resistant gene and the sensitive gene.

[0131] The embodiments of the present invention can verify the accuracy of the model, avoid erroneous decisions of the model, and improve the credibility of the model by identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-resistance association characteristics. The model reliability refers to the accuracy, stability, and consistency of a model in predicting or explaining a phenomenon under specific conditions.

[0132] As an embodiment of the present invention, the method of identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association feature includes: identifying the sensitivity gene corresponding to the drug-target interaction relationship, and identifying the drug resistance gene corresponding to the gene-drug resistance association feature; collecting association data of the sensitivity gene and the drug resistance gene, and classifying and labeling the association data according to the drug-target interaction relationship and the gene-drug resistance association feature to obtain classification label data; performing data verification on the drug resistance analysis model based on the classification label data to obtain a data verification result; defining an evaluation index of the drug resistance analysis model according to the data verification result; calculating the accuracy of the drug resistance analysis model based on the evaluation index; and identifying the model reliability of the drug resistance analysis model according to the accuracy.

[0133] Among them, the associated data refers to the data set related to sensitive genes and resistance genes, the classified labeled data refers to the data set after the feature data is classified and labeled, the data verification result refers to the result obtained after testing the resistance analysis model using the classified labeled data, the accuracy rate refers to the proportion of correctly predicted samples to the total number of samples, and the evaluation index refers to the evaluation standard for measuring model performance.

[0134] Optionally, the data verification of the drug resistance analysis model based on the classification labeling data can be obtained through cross-validation, and the classification labeling of the associated data based on the drug-target interaction relationship and the gene-resistance association characteristics can be achieved using a supervised learning method of a machine learning model.

[0135] In an optional embodiment of the present invention, based on the evaluation index, the accuracy of the drug resistance analysis model is calculated using the following formula:

[0136]

[0137] Among them, H represents the accuracy of the drug resistance analysis model, F represents the number of samples for which the drug resistance analysis model correctly predicts the drug resistance results, and T represents the number of samples for which the drug resistance analysis model incorrectly predicts the drug resistance results.

[0138] S6. Based on the anti-tumor drug resistance knowledge graph, define the result interpretation of the drug resistance analysis model, and output the anti-tumor drug resistance detection result of the tumor patient in combination with the reliability of the model and the result interpretation.

[0139] By defining the interpretation of the results of the drug resistance analysis model based on the anti-tumor drug resistance knowledge graph, the embodiment of the present invention can ensure that the output of the model can be correctly understood to guide the personalized treatment and drug selection of tumor diseases. The result interpretation refers to the explanation and elaboration of the model output results.

[0140] Optionally, based on the anti-tumor drug resistance knowledge graph, the result interpretation definition of the drug resistance analysis model can be implemented using a model explanatory analysis tool, such as a SHAP value tool.

[0141] Furthermore, the embodiment of the present invention combines the reliability of the model and the interpretation of the results to output the anti-tumor resistance test results of the tumor patient, which can guide doctors to select more suitable drugs for patients, thereby improving the treatment effect, reducing the physical and economic burden of ineffective treatment, and improving the quality of life of patients. At the same time, it can also help relevant personnel design new drugs that can overcome drug resistance, thereby promoting the advancement of anti-tumor drugs. The anti-tumor resistance test results refer to data and conclusions on the degree of resistance of tumor cells to specific anti-tumor drugs obtained through a series of experiments and analysis methods, such as the sensitivity of tumor cells to specific drugs.

[0142] Compared with the problems described in the background technology, the embodiment of the present invention can extract tumor molecular markers of the tumor cells by identifying the tumor cells of the tumor patient, collecting the genomic data of the tumor cells, and analyzing the genetic changes of the tumor cells, so as to help understand the molecular mechanism of tumor cell resistance and the heterogeneity within the tumor, and avoid the use of drugs that may induce drug resistance; further, the embodiment of the present invention can enhance the richness and accuracy of the anti-tumor drug resistance knowledge graph by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers, so as to better reveal the molecular mechanism of tumor resistance; in addition, the embodiment of the present invention can extract the gene-protein characteristics of the tumor cells based on the genetic changes, which can be used as input data for predicting tumor resistance models, monitor tumor progression and the development of drug resistance, and predict the effects of drugs with different chemical properties on specific tumor cell lines by identifying the drug chemical properties of the anti-tumor drugs; secondly, the embodiment of the present invention can predict the effects of drugs with different chemical properties on specific tumor cell lines by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers. The fusion features are used to construct a drug resistance analysis model for the tumor cells, which can dynamically monitor changes in tumor cell resistance and predict tumor cell resistance to specific anti-tumor drugs, thereby helping to screen and develop new anti-tumor drugs. Thirdly, the embodiments of the present invention analyze the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph and extract the gene-drug resistance association features of the tumor cells. This can predict the patient's possible response and efficacy to specific drugs, reveal the potential mechanism of tumor cell resistance, and identify the key genes that cause tumor cells to develop drug resistance, so as to verify the accuracy of the model, avoid incorrect decisions of the model, and improve the credibility of the model. Finally, the embodiments of the present invention output the anti-tumor drug resistance test results of the tumor patient by combining the reliability of the model and the interpretation of the results. This can guide doctors to select more suitable drugs for patients, thereby improving treatment effects, reducing the physical and economic burden of ineffective treatment, and improving the quality of life of patients. It can also help relevant personnel design new drugs that can overcome drug resistance, thereby promoting the advancement of anti-tumor drugs. Therefore, the artificial intelligence-based anti-tumor drug resistance detection method and system provided in the embodiments of the present invention can improve the accuracy of anti-tumor drug resistance detection.

[0143] Example 2:

[0144] like Figure 2 The figure shows a functional module diagram of an anti-tumor drug resistance detection system based on artificial intelligence of the present invention.

[0145] The AI-based anti-tumor drug resistance detection system 200 described in the present invention can be installed in an electronic device. Depending on the functionality implemented, the AI-based anti-tumor drug resistance detection system can include a gene analysis module 201, a knowledge graph construction module 202, an associated feature extraction module 203, a model generation module 204, a model verification module 205, and a result output module 206. A module, also referred to as a unit, is a series of computer program segments that can be executed by an electronic device processor and perform a fixed function, and is stored in the electronic device's memory.

[0146] In the embodiment of the present invention, the functions of each module / unit are as follows:

[0147] The gene analysis module 201 is used to obtain the anti-tumor drug to be tested and its corresponding tumor patient, identify the tumor cells of the tumor patient, extract tumor molecular markers of the tumor cells, collect genomic data of the tumor cells, and analyze the genetic changes of the tumor cells based on the genomic data;

[0148] The knowledge graph construction module 202 is used to collect the drug structure data and drug target data of the anti-tumor drug, and collect the drug sensitivity data of the tumor cells, and combine the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers to construct the anti-tumor resistance knowledge graph of the tumor patient.

[0149] The correlation feature extraction module 203 is used to extract the gene-protein feature of the tumor cell based on the gene change, identify the drug chemical properties of the anti-tumor drug according to the drug structure data, and calculate the degree of correlation between the gene-protein feature and the drug chemical properties;

[0150] A model generation module 204 is configured to perform dimensionality reduction fusion processing on the gene-protein features and the drug chemical properties according to the degree of association to obtain a fusion feature, and construct a drug resistance analysis model for the tumor cells based on the fusion feature;

[0151] a model validation module 205 for analyzing the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph, extracting the gene-drug resistance association characteristics of the tumor cells, and identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association characteristics;

[0152] The result output module 206 is used to define the result interpretation of the drug resistance analysis model according to the anti-tumor drug resistance knowledge graph, and output the anti-tumor drug resistance detection result of the tumor patient in combination with the model reliability and the result interpretation.

[0153] In detail, each module in the anti-tumor drug resistance detection system 200 based on artificial intelligence in the embodiment of the present invention adopts the same method as above when in use. Figure 1 The same technical means are used as the artificial intelligence-based anti-tumor resistance detection method described in , and can produce the same technical effects, so I will not go into details here.

[0154] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An artificial intelligence-based anti-tumor drug resistance detection method, characterized in that: The method comprises: Obtaining an anti-tumor drug to be tested and its corresponding tumor patient, identifying tumor cells of the tumor patient, extracting tumor molecular markers of the tumor cells, collecting genomic data of the tumor cells, and analyzing genetic changes in the tumor cells based on the genomic data; Collecting drug structure data and drug target data of the anti-tumor drug, and collecting drug sensitivity data of the tumor cells, and combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular markers to construct a knowledge graph of anti-tumor drug resistance of the tumor patient; extracting gene-protein characteristics of the tumor cells based on the gene changes, identifying the pharmacochemical properties of the anti-tumor drug based on the drug structure data, and calculating the degree of correlation between the gene-protein characteristics and the pharmacochemical properties; According to the degree of association, the gene-protein feature and the drug chemical property are subjected to fusion dimensionality reduction processing to obtain a fusion feature, and based on the fusion feature, a drug resistance analysis model of the tumor cell is constructed; Based on the anti-tumor drug resistance knowledge graph, the drug-target interaction relationship of the tumor patient is analyzed, and the gene-drug resistance association characteristics of the tumor cells are extracted. According to the drug-target interaction relationship and the gene-drug resistance association characteristics, the model reliability of the drug resistance analysis model is identified; According to the anti-tumor drug resistance knowledge graph, the result interpretation of the drug resistance analysis model is defined, and the anti-tumor drug resistance detection result of the tumor patient is output in combination with the reliability of the model and the result interpretation.

2. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: Analyzing the genetic changes of the tumor cells based on the genomic data includes: extracting tumor suppressor genes and viral oncogenes of the tumor cells based on the genomic data; Detecting the functional impairment of the tumor suppressor gene and detecting the abnormal activation state of the viral oncogene; Analyzing changes in expression levels of the tumor suppressor gene and the viral oncogene based on the functional impairment manifestations and the abnormal activation state; Identifying factors influencing the changes in expression levels and collecting molecular genetic data corresponding to the influencing factors; The gene changes of the tumor cells are analyzed by combining the expression level changes and the molecular genetics data.

3. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The step of combining the drug structure data, the drug target data, the drug sensitivity data, and the tumor molecular markers to construct a knowledge graph of anti-tumor drug resistance of the tumor patient includes: Determine the disease type of the tumor patient and its corresponding field range by combining the drug structure data, the drug target data, the drug sensitivity data and the tumor molecular marker; Based on the disease type, extracting the drug structure data, the drug target data, the drug sensitivity data and the core keywords of the tumor molecular markers; Extracting entity categories corresponding to the disease types based on the core keywords; Based on the domain scope, collecting domain knowledge of the disease type; Identifying the relationship patterns between the entity categories based on the domain knowledge and extracting attribute features of the entity categories; Based on the entity categories, the relationship patterns and the attribute characteristics, a knowledge graph of the anti-tumor drug resistance of the tumor patients is constructed.

4. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The extracting of gene-protein characteristics of the tumor cells based on the gene changes includes: Based on the gene changes, identifying the mutated genes of the tumor cells; Extracting the encoded protein corresponding to the mutated gene and analyzing the status of the encoded protein; Determining, based on the status manifestations, a gene-protein functional association relationship between the mutated gene and the protein encoding the protein; Based on the gene-protein functional association relationship, constructing a gene-protein relationship network of the mutated gene and the protein encoding the protein; The gene-protein characteristics of the tumor cells are extracted based on the gene-protein relationship network.

5. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: According to the degree of association, the gene-protein feature and the drug chemical property are subjected to fusion dimensionality reduction processing to obtain a fusion feature, including: Collecting correlation data between the gene-protein feature and the drug chemical property according to the correlation degree, and extracting correlation features between the gene-protein feature and the drug chemical property; Performing data integration processing on the associated data to obtain integrated data; Based on the association features, identifying the disease type of the gene-protein feature, and analyzing the drug response of the disease type; extracting key features of the gene-protein signature and the drug chemical properties from the integrated data according to the disease type and the drug response; Execute the key feature dimensionality reduction processing to obtain the reduced dimensionality feature; Based on the integrated data and the dimensionality reduction features, a fusion feature of the gene-protein feature and the drug chemical property is obtained.

6. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The method of constructing a drug resistance analysis model of the tumor cells based on the fusion characteristics comprises: Identifying a tumor patient corresponding to the tumor cell, and analyzing the disease type of the tumor patient based on the fusion signature; extracting biomarker characteristics of the tumor cells according to the disease type; Based on the biomarker characteristics, identifying factors affecting drug resistance of the tumor cells; Analyzing the potential drug resistance mechanism of the tumor cells based on the factors affecting drug resistance; Combining the disease type, the drug resistance influencing factors and the potential drug resistance mechanism, a drug resistance analysis model for the tumor cells is constructed.

7. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The analyzing the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph includes: Based on the anti-tumor drug resistance knowledge graph, identifying the anti-cancer drugs for the tumor patient and their corresponding target proteins; extracting structural features of the anticancer drug and the target protein; Identifying the binding site between the anticancer drug and the target protein based on the structural characteristics; Based on the binding site, performing molecular docking processing between the anticancer drug and the target protein to obtain a molecular docking result; Analyzing the binding mode between the anticancer drug and the target protein according to the molecular docking results; Based on the binding mode and the molecular docking results, identifying the protein function of the target protein; Combining the protein function and the binding mode, the drug-target interaction relationship of the tumor patient is analyzed.

8. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The extracting the gene-drug resistance association features of the tumor cells based on the anti-tumor drug resistance knowledge graph includes: Extracting drug resistance genes and sensitivity genes of the tumor cells based on the anti-tumor drug resistance knowledge graph; Analyzing the gene expression levels of the drug-resistant gene and the drug-sensitive gene; Calculating the difference coefficient between the drug-resistant gene and the drug-sensitive gene according to the gene expression levels; extracting the gene characteristics of the drug resistance gene based on the difference coefficient; analyzing the drug resistance formation mechanism of the tumor cells based on the gene characteristics; Identifying the causes of the resistance development mechanism; Based on the causes, analyzing the drug resistance change patterns of the tumor cells; Combining the drug resistance formation mechanism and the drug resistance change pattern, the gene-drug resistance association characteristics of the tumor cells are extracted.

9. The method for detecting anti-tumor drug resistance based on artificial intelligence according to claim 1, wherein: The step of identifying the reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association characteristics includes: Identifying the sensitivity genes corresponding to the drug-target interaction relationship, and identifying the drug resistance genes corresponding to the gene-drug resistance association characteristics; Collecting association data of the sensitive genes and the drug-resistant genes, and classifying and labeling the association data according to the drug-target interaction relationship and the gene-drug resistance association characteristics to obtain classification label data; Based on the classification mark data, data verification is performed on the drug resistance analysis model to obtain a data verification result; Defining evaluation indicators of the drug resistance analysis model according to the data verification results; Calculating the accuracy of the drug resistance analysis model based on the evaluation index; According to the accuracy rate, the model reliability of the drug resistance analysis model is identified.

10. An artificial intelligence-based anti-tumor drug resistance detection system, characterized in that: The system comprises: A gene analysis module is used to obtain the anti-tumor drug to be tested and its corresponding tumor patient, identify the tumor cells of the tumor patient, extract tumor molecular markers of the tumor cells, collect genomic data of the tumor cells, and analyze the genetic changes of the tumor cells based on the genomic data; a knowledge graph construction module, configured to collect drug structure data and drug target data of the anti-tumor drug, and drug sensitivity data of the tumor cells, and to construct a knowledge graph of anti-tumor drug resistance of the tumor patient by combining the drug structure data, the drug target data, the drug sensitivity data, and the tumor molecular markers; a correlation feature extraction module, configured to extract the gene-protein features of the tumor cells based on the gene changes, identify the pharmacochemical properties of the anti-tumor drug based on the drug structure data, and calculate the degree of correlation between the gene-protein features and the pharmacochemical properties; a model generation module, configured to perform dimensionality reduction fusion processing on the gene-protein features and the drug chemical properties according to the degree of association to obtain a fusion feature, and construct a drug resistance analysis model for the tumor cells based on the fusion feature; a model validation module for analyzing the drug-target interaction relationship of the tumor patient based on the anti-tumor drug resistance knowledge graph, extracting the gene-drug resistance association characteristics of the tumor cells, and identifying the model reliability of the drug resistance analysis model based on the drug-target interaction relationship and the gene-drug resistance association characteristics; The result output module is used to define the result interpretation of the resistance analysis model according to the anti-tumor resistance knowledge graph, and output the anti-tumor resistance detection result of the tumor patient in combination with the reliability of the model and the result interpretation.

Images