Drug recommendation method and device, computer equipment and storage medium

Abstract

The invention belongs to the field of artificial intelligence and digital medical treatment, and relates to a drug recommendation method, which comprises the following steps: preprocessing medical record data and gene data according to a preset data standard specification to obtain standard medical record data and standard gene data, and fusing the standard medical record data and the standard gene data through a health analysis model for prediction to obtain a drug recommendation result. And analyzing the health analysis result, the standard medical record data and the standard gene data through a drug recommendation model to obtain a drug recommendation scheme. The invention further provides a drug recommendation device, computer equipment and a storage medium. In addition, the invention also relates to a block chain technology, and the medical record data and the gene data can be stored in a block chain. According to the method, the efficiency and accuracy of drug recommendation can be improved, the personalized level of drug recommendation is remarkably improved, drug treatment is more targeted and effective, and the drug use safety of patients is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence and digital medical technology, and in particular to a drug recommendation method, device, computer equipment and storage medium. Background Art

[0002] With the development of medical technology and bioinformatics, personalized medicine has gradually become an important means to improve treatment effects and patient satisfaction. Traditional drug recommendations mainly rely on doctors' experience and existing clinical guidelines, but this method is difficult to fully consider the unique situation of each patient. Genetic data and medical records can provide rich individual information, but medical record data and genetic data usually have heterogeneity and non-standardization problems. Due to the complexity and diversity of these data, the information dimensions contained vary greatly, making it difficult to directly apply them to subsequent drug recommendation tasks, which in turn affects the comprehensive characterization of the patient's health status, resulting in poor efficiency and accuracy of drug recommendations. Summary of the invention

[0003] The purpose of the embodiments of the present application is to propose a drug recommendation method, apparatus, computer device and storage medium to solve the technical problem of poor efficiency and accuracy of drug recommendation in the prior art.

[0004] In order to solve the above technical problems, the present application embodiment provides a drug recommendation method, which adopts the following technical solution:

[0005] Obtaining medical record data and gene data of a target patient, and preprocessing the medical record data and gene data according to preset data standard specifications to obtain standard medical record data and standard gene data;

[0006] Inputting the standard medical record data and the standard gene data into the trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain a health analysis result;

[0007] The health analysis results, the standard medical record data and the standard gene data are input into the trained drug recommendation model to perform drug screening and drug effect prediction, and a drug recommendation plan is output.

[0008] Furthermore, the health analysis model includes a feature extraction layer, a medical record analysis layer, a gene analysis layer and a fusion output layer, and the step of inputting the standard medical record data and the standard gene data into the trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain a health analysis result includes:

[0009] Inputting the standard medical record data and the standard gene data into the feature extraction layer to perform feature extraction, and obtaining corresponding medical record data features and gene data features;

[0010] The medical record analysis layer fuses the medical record data features and the gene data features to perform analysis and prediction to obtain a disease analysis result;

[0011] The gene data features are analyzed and predicted by the gene analysis layer to obtain gene variation and drug action results;

[0012] The disease analysis results and the gene variation and drug action results are input into the fusion output layer, and the health analysis results are obtained through a data fusion algorithm.

[0013] Furthermore, the gene analysis layer includes a gene identification sublayer and a drug efficacy analysis prediction sublayer, and the step of analyzing and predicting the gene data features through the gene analysis layer to obtain gene variation and drug effect results includes:

[0014] Comparing the gene sequences in the gene data features and identifying the mutation sites through the gene identification sublayer to obtain gene mutation information;

[0015] According to the gene variation information, obtaining drug response information corresponding to the gene variation information by querying a pre-constructed gene variation and drug response association database;

[0016] The gene variation information and the drug response information are input into the drug efficacy analysis prediction sublayer, and the gene variation information and the drug response information are jointly analyzed through a pre-trained large language model to obtain the gene variation and drug action results.

[0017] Furthermore, the fusion output layer includes a feature fusion sublayer and a health prediction sublayer, and the step of inputting the disease analysis result and the gene variation and drug action result into the fusion output layer and obtaining the health analysis result through the data fusion algorithm includes:

[0018] The feature fusion sublayer uses multidimensional data fusion technology to fuse the disease analysis results and the gene variation and drug action results to obtain fusion features;

[0019] Through the health prediction sublayer, individual feature recognition and health analysis are performed on the fusion features to obtain health analysis results.

[0020] Furthermore, the drug recommendation model includes a drug screening layer, a drug effect prediction layer, a recommendation result generation layer and an explanation layer. The steps of inputting the health analysis results, the standard medical record data and the standard gene data into the trained drug recommendation model for drug screening and drug effect prediction, and outputting a drug recommendation plan include:

[0021] The drug screening layer extracts key information and represents semantic vectors from the health analysis results, the standard medical record data, and the standard gene data to obtain drug semantic features, and obtains candidate drugs through the drug semantic features;

[0022] By combining the candidate drugs with the drug effect prediction layer, the key information of drug effects is extracted from the health analysis results, the standard medical record data and the standard gene data to obtain drug efficacy and adverse reaction results;

[0023] The recommendation result generation layer combines the drug efficacy and adverse reaction results to score the candidate drugs and screen out the optimal drug combination;

[0024] The optimal drug combination is annotated with drug usage instructions through the explanation layer to obtain a drug recommendation plan.

[0025] Furthermore, the step of extracting key information of drug effects from the health analysis results, the standard medical record data and the standard gene data by combining the drug effect prediction layer with the candidate drug to obtain drug efficacy and adverse reaction results includes:

[0026] In the drug effect prediction layer, a condition semantic vector is obtained by extracting key features from the health analysis results, the standard medical record data and the standard gene data;

[0027] Compare the standard gene data with the gene database to obtain SNP information of key gene sites, and obtain a drug effect feature vector based on the SNP information and the candidate drug;

[0028] The disease semantic vector and the drug effect feature vector are feature fused to obtain drug efficacy and adverse reaction results.

[0029] Furthermore, after the step of outputting the drug recommendation scheme, the method further includes:

[0030] Collecting feedback data from the target patient after using the recommended drug regimen, wherein the feedback data includes drug efficacy and adverse drug reactions;

[0031] The health analysis model and the drug recommendation model are updated according to the feedback data.

[0032] In order to solve the above technical problems, the present application also provides a drug recommendation device, which adopts the following technical solution:

[0033] An acquisition module is used to acquire the medical record data and gene data of a target patient, and pre-process the medical record data and gene data according to a preset data standard specification to obtain standard medical record data and standard gene data;

[0034] A health analysis module, used for inputting the standard medical record data and the standard gene data into a trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain a health analysis result;

[0035] The drug recommendation module is used to input the health analysis results, the standard medical record data and the standard gene data into the trained drug recommendation model to perform drug screening and drug effect prediction, and output a drug recommendation plan.

[0036] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0037] The computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the drug recommendation method described above when executing the computer-readable instructions.

[0038] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0039] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the drug recommendation method described above are implemented.

[0040] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0041] The present application provides a drug recommendation method, which obtains standard medical record data and standard gene data by preprocessing medical record data and gene data according to preset data standard specifications, thereby reducing data noise and improving data quality. At the same time, the consistency of the data can speed up the calculation and improve the interpretability of data analysis. By fusing standard medical record data and standard gene data with a health analysis model for prediction, multi-dimensional data in different feature spaces can be extracted, and the characteristics of the target patient can be fully acquired. At the same time, data from different sources can be fused to eliminate redundancy and noise, thereby improving the accuracy and reliability of the analysis. By analyzing the health analysis results, standard medical record data and standard gene data with a drug recommendation model, the efficiency and accuracy of drug recommendations can be improved, the personalization level of drug recommendations can be significantly improved, drug treatment can be more targeted and effective, and the safety of drug use for patients can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0044] Figure 2 is a flow chart of an embodiment of a drug recommendation method according to the present application;

[0045] Figure 3 yes Figure 2 A flowchart of a specific implementation of step S202;

[0046] Figure 4 is a schematic structural diagram of an embodiment of a drug recommendation device according to the present application;

[0047] Figure 5 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0049] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0051] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0052] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0053] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0054] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0055] It should be noted that the drug recommendation method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the drug recommendation device is generally set in the server / terminal device.

[0056] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0057] Continue to refer Figure 2 , shows a flow chart of an embodiment of a drug recommendation method according to the present application, comprising the following steps:

[0058] Step S201, obtaining the medical record data and gene data of the target patient, preprocessing the medical record data and gene data according to preset data standard specifications, and obtaining standard medical record data and standard gene data.

[0059] In this embodiment, the electronic device (eg, Figure 1 The server / terminal device shown in the figure can obtain the medical record data and gene data of the target patient through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.

[0060] According to the target patient's identity information, the target patient's detailed medical record data is obtained from the hospital's electronic medical record system, including past medical history, symptoms, diagnosis records, treatment records, drug use records, etc. Gene data can be obtained from a standard gene database based on the target patient's identity information, or the target patient's whole genome or specific gene sites can be sequenced through high-throughput sequencing technology based on the target patient's gene sequencing data to obtain the corresponding gene data. Among them, the gene data includes amino acid variation data, gene variation data, copy number variation (CNVs) and fused structural variation, and the amino acid variation data includes single nucleotide polymorphisms (SNPs), synonymous mutations, missense mutations, nonsense mutations, and stop codon mutations.

[0061] The acquired medical record data and gene data are preprocessed according to preset data standard specifications, including: for the medical record data and gene data, a data cleaning algorithm is used to remove noise, outliers, missing values, etc. in the data to obtain cleaned medical record data and gene data. For example, the Pandas library in Python is used to clean the data, remove features with missing values ​​exceeding 20%, and fill the remaining missing values ​​with the mean; according to the preset data standard specifications, the cleaned medical record data and gene data are converted to convert unstructured data into structured data, and the data format is unified to obtain converted medical record data and gene data; for the converted medical record data and gene data, a data standardization algorithm is used to normalize the data to eliminate differences in data dimension and data scale to obtain standardized medical record data and gene data, that is, standard medical record data and standard gene data.

[0062] Specifically, an outlier detection algorithm, such as the distance-based LOF algorithm, is used to remove noise and outliers in the data, and the maximum likelihood estimation method is used to interpolate missing values ​​to obtain cleaned medical record data and gene data. The cleaned medical record data and gene data are converted to convert the unstructured medical record data into a structured XML format and the gene data into a FASTA format, and the converted medical record data and gene data are obtained accordingly. For the converted medical record data and gene data, the minimum-maximum normalization algorithm is used to map the data to the [0,1] interval, eliminate the differences in data dimension and data scale, and obtain standardized medical record data and gene data.

[0063] It should be emphasized that in order to further ensure the privacy and security of medical record data and genetic data, the above medical record data and genetic data can also be stored in a blockchain node.

[0064] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0065] Step S202, inputting the standard medical record data and the standard gene data into the trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain the health analysis result.

[0066] In this embodiment, the trained health analysis model is used to extract features from the standard medical record data and the standard gene data, respectively, to obtain medical record data features that characterize the clinical characteristics of the target patient and gene data features of key gene sites and gene expression patterns related to the health status of the target patient; the medical record data features and the gene data features are spliced ​​and fused to obtain health features that characterize the health of the target patient; by analyzing key information in the health features, such as disease type, course of disease, severity, etc., the condition analysis results of the target patient are obtained, and the condition analysis results include clinical symptoms and signs, laboratory test results, treatment response data (including drug usage, dosage, treatment duration), treatment effect and treatment needs, etc.; Based on the characteristics, gene sequence alignment and variation detection are performed to identify disease-related gene variations. Based on known gene-drug interaction databases (such as PharmGKB and DrugBank), the effects of gene variations on drug response are analyzed to obtain drug response data. By combining gene variation and drug response data, the model reasoning ability is used to predict the target patient's sensitivity to different drugs and potential adverse reactions, namely, gene variation and drug action results. Finally, a large language model is used to integrate the disease analysis results and gene variation and drug action results to identify the target patient's overall health status and individual characteristics, namely, health analysis results. The health analysis results are used to illustrate potential health problems that may exist in the target patient in order to provide early warning information.

[0067] In some optional implementations, the health analysis model includes a feature extraction layer, a medical record analysis layer, a gene analysis layer, and a fusion output layer, see Figure 3 As shown, the steps of inputting the standard medical record data and the standard gene data into the trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain the health analysis result include:

[0068] Step S301, inputting the standard medical record data and the standard gene data into the feature extraction layer for feature extraction to obtain corresponding medical record data features and gene data features.

[0069] Among them, the feature extraction layer includes a medical record feature extraction sublayer for extracting features from medical record data and a gene feature extraction sublayer for extracting features from gene data. The standard medical record data is input into the medical record feature extraction sublayer, and the standard gene data is input into the gene feature extraction sublayer for feature extraction to obtain corresponding medical record data features and gene data features.

[0070] Specifically, the medical record feature extraction sublayer adopts a BERT model pre-trained based on data in the medical field, such as the BioBERT model or the ClinicalBERT model. For standard medical record data, it uses named entity recognition technology combined with medical dictionaries and knowledge graphs (such as UMLS, SNOMED CT) to identify and label key medical information such as symptoms, signs, test results, and condition descriptions. By analyzing and extracting the labeled results, the specific content and attribute information of entities such as symptoms, signs, test results, and condition descriptions are obtained, and the medical record data features can be obtained by output.

[0071] The gene feature extraction sublayer uses a sequence autoencoder based on a recursive neural network model, such as a long short-term memory network (LSTM) or a bidirectional long short-term memory network (Bi-LSTM), to extract gene data features through recursion and attention mechanisms.

[0072] Step S302, the medical record data features and the gene data features are integrated through the medical record analysis layer to perform analysis and prediction to obtain the disease analysis result.

[0073] In this embodiment, the medical record analysis layer includes a feature fusion sublayer and an analysis prediction sublayer. The feature fusion sublayer uses a bidirectional long short-term memory network (Bi-LSTM) to capture the temporal dependency between medical record data features and gene data features through LSTM units in the forward and backward directions to obtain a medical record analysis vector. The condition analysis prediction sublayer uses a pre-trained large language model (such as GPT-4, BERT) to analyze the medical record analysis vector, predict the target patient's condition development trend, and output the target patient's condition analysis results.

[0074] Step S303, analyzing and predicting the gene data features through the gene analysis layer to obtain gene variation and drug action results.

[0075] Among them, the gene analysis layer includes a gene recognition sublayer and a pharmacodynamic analysis prediction sublayer. Specifically, the gene recognition sublayer compares the gene sequences in the gene data features and identifies the mutation sites to obtain gene mutation information; based on the gene mutation information, the drug reaction information corresponding to the gene mutation information is obtained by querying a pre-constructed gene mutation and drug reaction association database; the gene mutation information and drug reaction information are input into the pharmacodynamic analysis prediction sublayer, and the gene mutation information and drug reaction information are jointly analyzed through a pre-trained large language model to obtain the gene mutation and drug action results.

[0076] In this embodiment, the gene identification sublayer can use the GWAS analysis (Genome wide association study, whole genome association analysis) method to perform differential expression analysis at the whole genome level to determine the key SNP sites and disease-related genes in the gene data characteristics that are related to the health status of the target patient, and obtain gene variation information.

[0077] Specifically, the BWA tool in the GWAS analysis method is used to align the sequencing reads in the gene data features to the reference genome, and the GATK tool in the GWAS analysis method is used to detect mutations, identify the gene mutations carried by the target patients, and obtain gene mutation information. According to the pre-constructed database containing 500 common gene variants and drug responses, the drug response information corresponding to the gene variant information is obtained. Among them, the gene variant and drug response association database refers to a database that collects and integrates gene variant information and drug response data. These databases usually contain detailed information on how gene variants affect the pharmacokinetics and pharmacodynamics of drugs.

[0078] The drug efficacy analysis prediction sublayer uses a pre-trained large language model, which is trained based on the collected gene variation data and the corresponding drug response data. The drug response data includes the pharmacokinetic and pharmacodynamic properties of the drug. During the training process, the model is continuously trained to identify the patterns and associations between gene variation and drug response. The trained large language model can infer unknown gene variation data and predict its impact on drug response, that is, to obtain the results of gene variation and drug action.

[0079] By jointly analyzing gene variation information and drug response information, combining gene variation with drug response can help doctors choose drugs that are more likely to be effective in patients and avoid using drugs that may cause serious adverse reactions. It can also predict adverse reactions, take preventive measures in advance, and reduce patient risks, thereby reducing medical costs and improving treatment efficiency.

[0080] Step S304, input the disease analysis results and the gene variation and drug action results into the fusion output layer, and obtain the health analysis results through the data fusion algorithm.

[0081] In this embodiment, the fusion output layer includes a feature fusion sublayer and a health prediction sublayer. The feature fusion sublayer uses multidimensional data fusion technology to fuse the disease analysis results and gene mutations with the drug action results to obtain fusion features; the health prediction sublayer performs individual feature recognition and health analysis on the fusion features to obtain health analysis results.

[0082] Among them, the multidimensional data fusion technology can adopt convolutional neural network (CNN), encoder fusion method and attention mechanism-based fusion method, among which the encoder in the encoder fusion method can use BERT encoder or LSTM encoder.

[0083] The health prediction sublayer can use the BERT series models to analyze the fusion features and extract key features related to health prediction. These key features can reflect the health status of the individual; and analyze these key features to form a comprehensive health profile of the target patient. Based on the comprehensive health profile, the health status of the target patient is predicted in real time to obtain the health analysis results. The health analysis results include the overall health status and individual health status of the target patient. The overall health status can be a health score, disease risk prediction or other forms of analysis results. The individual health status is a personalized health indicator that characterizes the health status of the target patient, including physiological health indicators, living habits, nutritional status, chronic disease status, genetic and family medical history, immune status, drug sensitivity, etc.

[0084] In some optional implementations of this embodiment, the health analysis results are generated into a health analysis report and pushed to doctors and patients through the hospital's mobile APP. Doctors can view the patient's health status change curve and important gene site information to provide a reference for subsequent treatment plans; patients can view their own health risk level and improvement suggestions, adjust their daily diet and living habits, and improve their health awareness.

[0085] By integrating the results of disease analysis with genetic variation and drug action results, it is possible to comprehensively analyze information from different data sources and dimensions, provide a more comprehensive health status assessment, improve the accuracy of diagnosis, help develop personalized treatment plans, and better manage patients' health conditions; it is able to predict possible diseases or health problems in the future, thereby achieving early intervention, which can reduce the need for expensive treatments and long-term care, thereby reducing overall medical costs.

[0086] This application uses a health analysis model to fuse standard medical record data and standard genetic data for prediction, which can extract multi-dimensional data from different feature spaces, obtain characteristics that comprehensively characterize the target patients, and at the same time fuse data from different sources to eliminate redundancy and noise, thereby improving the accuracy and reliability of health analysis results.

[0087] Step S203, input the health analysis results, standard medical record data and standard gene data into the trained drug recommendation model to perform drug screening and drug effect prediction, and output a drug recommendation plan.

[0088] In this embodiment, the drug recommendation model is based on a large language model, and uses natural language processing technology and deep learning algorithms to perform in-depth analysis of health analysis results, standard medical record data, and standard genetic data to obtain a drug recommendation plan.

[0089] In some optional implementations, the drug recommendation model includes a drug screening layer, a drug effect prediction layer, a recommendation result generation layer, and an explanation layer. The steps of inputting the health analysis results, standard medical record data, and standard gene data into the trained drug recommendation model for drug screening and drug effect prediction and outputting a drug recommendation plan include:

[0090] Through the drug screening layer, key information is extracted and semantic vectors are expressed from health analysis results, standard medical record data, and standard gene data to obtain drug semantic features, and candidate drugs are obtained through drug semantic features;

[0091] By combining candidate drugs through the drug effect prediction layer, key information on drug effects is extracted from health analysis results, standard medical record data, and standard gene data to obtain drug efficacy and adverse reaction results;

[0092] The recommendation result generation layer combines drug efficacy and adverse reaction results to score candidate drugs and screen out the optimal drug combination;

[0093] The explanation layer is used to mark the instructions for use of the optimal drug combination and obtain a recommended drug plan.

[0094] In this embodiment, the drug screening layer is used to extract drug semantic features related to the disease based on health analysis results, standard medical record data and standard gene data, and screen matching candidate drugs from existing drug databases (such as DrugBank, PharmGKB, and RxNorm) based on the drug semantic features. Drug databases are online resources for storing and retrieving drug-related information, which usually contain detailed information such as drug name, drug chemical structure, pharmacological action, indication, side effect, dosage, drug interaction, and drug category.

[0095] The drug effect prediction layer uses a large language model and machine learning algorithms to predict drug efficacy and adverse drug reactions based on candidate drugs, health analysis results, standard medical record data, and standard gene data. Among them, drug efficacy includes therapeutic effect, therapeutic range, onset time, duration of efficacy, efficacy intensity, efficacy stability, etc.; adverse drug reactions include side effects, toxic effects, sequelae, allergic reactions, specific reactions, and drug interactions. In this embodiment, drug efficacy prediction can be predicted using an intelligent decision-making algorithm, and drug adverse reaction prediction can be predicted using a drug adverse reaction prediction model, which is trained based on an adverse drug reaction database (such as SIDER).

[0096] The recommendation result generation layer combines the drug efficacy and adverse reaction results to comprehensively score each candidate drug, obtain the scoring results, select the optimal drug combination according to the scoring results, and ensure the effectiveness and safety of the recommended solution. For example, determine the scoring criteria and weights of various indicators such as drug efficacy, adverse drug reactions, and drug interactions, where the weight distribution should be determined based on clinical importance and treatment goals; evaluate the drug efficacy based on the drug efficacy data, such as clinical trial results, drug concentration-effect relationship, etc., and convert it into a quantitative score to obtain the drug efficacy score; quantitatively score the adverse reaction data, including the type, frequency, and severity of the adverse reaction, to obtain the adverse reaction score; evaluate the possible interactions between candidate drugs or between candidate drugs and food, as well as the potential impact of these interactions on patient health, and score them to obtain the drug action score; based on the corresponding weights, the drug efficacy score, adverse reaction score, and drug action score are weighted and summed to obtain the scoring result of each candidate drug.

[0097] The explanation layer can use a large language model to annotate the drug usage instructions for the optimal drug combination, and generate detailed drug usage instructions and potential risk warnings for drug usage. The drug usage instructions include the drug name, dosage, usage method, precautions, etc.; the potential risk warnings for drug usage include possible adverse reactions, drug interactions, etc.

[0098] The generated drug recommendation plan is displayed visually to facilitate doctors and patients to understand and use the recommendation plan. Detailed information and real-time consultation functions of drug recommendations can also be provided through an interactive interface.

[0099] By generating drug recommendation plans through drug recommendation models, personalized treatment can be provided based on the patient's specific condition, genotype, past treatment response and other personalized factors, making drug treatment more targeted and effective; by accurately matching drugs and patients, the treatment effect can be improved, ineffective treatment and drug waste can be reduced, thereby reducing medical costs; at the same time, the occurrence of adverse reactions can be reduced, and the quality and efficiency of overall medical services can be improved.

[0100] In some optional implementations of this embodiment, the above-mentioned steps of extracting key information of drug effects from health analysis results, standard medical record data and standard gene data by combining candidate drugs through the drug effect prediction layer to obtain drug efficacy and adverse reaction results include:

[0101] In the drug effect prediction layer, the semantic vector of the disease condition is obtained by extracting key features from health analysis results, standard medical record data, and standard gene data;

[0102] Compare the standard gene data with the gene database to obtain the SNP information of key gene sites, and obtain the drug effect feature vector based on the SNP information and candidate drugs;

[0103] The disease semantic vector and the drug effect feature vector are fused to obtain the drug efficacy and adverse reaction results.

[0104] Specifically, the key features such as medical history, symptoms, signs, auxiliary examinations, diagnosis, and treatment in the health analysis results, standard medical records, and standard gene data are extracted, and the key features are converted into semantic vectors through word embedding algorithms such as Word2Vec to obtain the semantic vector of the disease. At the same time, the standard gene data is compared with the standard gene database, and the SNP information of 50 key gene sites related to drug metabolism, drug targets, and adverse drug reactions is extracted, and converted into a drug effect feature vector through One-Hot encoding. Finally, the semantic vector of the disease and the drug effect feature vector are fused in a cascade manner to obtain a patient feature vector, which is input into a three-layer fully connected neural network model (FCNN). The fully connected neural network model includes an input layer, a hidden layer, and an output layer. The fully connected neural network model predicts the response of the target patient to each candidate drug according to the patient characteristics. The response includes the drug efficacy and adverse reaction risk of each candidate drug on the patient, that is, the drug efficacy and adverse reaction results are obtained.

[0105] By predicting drug efficacy and adverse reactions, we can provide patients with more personalized treatment plans, improve drug treatment effects and medication safety, and reduce the risk of adverse reactions.

[0106] This application obtains standard medical record data and standard gene data by preprocessing medical record data and gene data according to preset data standard specifications, which can reduce data noise and improve data quality. At the same time, the consistency of data can speed up the calculation and improve the interpretability of data analysis. By fusing standard medical record data and standard gene data through a health analysis model for prediction, multi-dimensional data in different feature spaces can be extracted to obtain characteristics that comprehensively characterize the target patient. At the same time, data from different sources can be fused to eliminate redundancy and noise, thereby improving the accuracy and reliability of the analysis. By analyzing the health analysis results, standard medical record data and standard gene data through a drug recommendation model, the level of personalization of drug recommendations can be significantly improved, making drug treatment more targeted and effective, and improving the safety of drug use for patients.

[0107] In some optional implementations, after the above step of outputting the drug recommendation scheme, the following steps are further included:

[0108] Collect feedback data from target patients after using the recommended drug regimen, including drug efficacy and adverse drug reactions;

[0109] Update health analysis models and drug recommendation models based on feedback data.

[0110] In this embodiment, after taking the medicine according to the recommended medicine plan, the patient will provide feedback data to the hospital in real time. Based on the feedback data, the health analysis model and the drug recommendation model are retrained and optimized to continuously improve the accuracy and personalization level of the recommendations.

[0111] In this embodiment, the patient's historical medical record data and historical genetic data are collected, the historical medical record data are preprocessed to obtain standard historical medical record data, the historical genetic data are preprocessed, encoded and standardized, and low-quality and repeated sequences are removed to obtain standard historical genetic data; natural language processing technologies such as regular expressions and bag-of-words models are used to extract key medical concepts and clinical events from the standard historical medical record data to obtain medical record data features, and structured representation is performed to construct the patient's electronic medical record knowledge graph; sequence alignment and variation detection are performed on the standard historical genetic data to identify mutation sites and variation types in the patient's genome to obtain genetic variation data; the electronic medical record knowledge graph and genetic variation data are integrated to construct health feature data for each patient; the pre-constructed health analysis model and drug recommendation model are trained using the health feature data to obtain the final health analysis model and drug recommendation model that have been trained and meet the requirements.

[0112] In order to verify whether the model meets the requirements, the three evaluation indicators commonly used in machine learning can be used to evaluate the model's prediction results: precision P (Precision), recall R (Recall) and F-score. When the precision P, recall R and F-score meet the preset thresholds, it means that the model meets the requirements; otherwise, the health analysis model and drug recommendation model are retrained.

[0113] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0114] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0115] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0116] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0117] Further references Figure 4 , as a response to the above Figure 2The present application provides an embodiment of a drug recommendation device, which is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0118] like Figure 4 As shown, the drug recommendation device 400 described in this embodiment includes: an acquisition module 401, a health analysis module 402 and a drug recommendation module 403. Among them:

[0119] The acquisition module 401 is used to acquire the medical record data and gene data of the target patient, and pre-process the medical record data and gene data according to the preset data standard specification to obtain standard medical record data and standard gene data;

[0120] The health analysis module 402 is used to input the standard medical record data and the standard gene data into the trained health analysis model, and obtain the health analysis result by fusing the standard medical record data and the standard gene data for prediction;

[0121] The drug recommendation module 403 is used to input the health analysis results, the standard medical record data and the standard gene data into the trained drug recommendation model to perform drug screening and drug effect prediction, and output a drug recommendation plan.

[0122] It should be emphasized that in order to further ensure the privacy and security of medical record data and genetic data, the above medical record data and genetic data can also be stored in a blockchain node.

[0123] Based on the above-mentioned drug recommendation device 400, by pre-processing the medical record data and gene data according to the preset data standard specifications, standard medical record data and standard gene data are obtained, which can reduce data noise and improve data quality. At the same time, the consistency of the data can speed up the calculation speed and improve the interpretability of data analysis; by fusing the standard medical record data and standard gene data through the health analysis model for prediction, multi-dimensional data in different feature spaces can be extracted to obtain comprehensive characteristics of the target patients. At the same time, data from different sources are fused to eliminate redundancy and noise, thereby improving the accuracy and reliability of the analysis; by analyzing the health analysis results, standard medical record data and standard gene data through the drug recommendation model, the personalization level of drug recommendations can be significantly improved, making drug treatment more targeted and effective, and improving the safety of patient medication.

[0124] In some optional implementations, the health analysis model includes a feature extraction layer, a medical record analysis layer, a gene analysis layer, and a fusion output layer, and the health analysis module 402 includes:

[0125] A feature extraction submodule, used for inputting the standard medical record data and the standard gene data into the feature extraction layer to perform feature extraction, and obtain corresponding medical record data features and gene data features;

[0126] The disease analysis submodule is used to analyze and predict the disease by fusing the medical record data features and the gene data features through the medical record analysis layer to obtain the disease analysis result;

[0127] A gene drug efficacy prediction submodule is used to analyze and predict the gene data characteristics through the gene analysis layer to obtain gene variation and drug effect results;

[0128] The fusion output submodule is used to input the disease analysis results and the gene variation and drug action results into the fusion output layer, and obtain the health analysis results through the data fusion algorithm.

[0129] By integrating standard medical record data and standard genetic data for prediction through the health analysis model, it is possible to extract multi-dimensional data from different feature spaces and obtain characteristics that comprehensively characterize the target patients. At the same time, data from different sources can be integrated to eliminate redundancy and noise, thereby improving the accuracy and reliability of health analysis results.

[0130] In some optional implementations of this embodiment, the gene analysis layer includes a gene identification sublayer and a drug efficacy analysis prediction sublayer, and the gene drug efficacy prediction submodule is further used to:

[0131] Comparing the gene sequences in the gene data features and identifying the mutation sites through the gene identification sublayer to obtain gene mutation information;

[0132] According to the gene variation information, obtaining drug response information corresponding to the gene variation information by querying a pre-constructed gene variation and drug response association database;

[0133] The gene variation information and the drug response information are input into the drug efficacy analysis prediction sublayer, and the gene variation information and the drug response information are jointly analyzed through a pre-trained large language model to obtain the gene variation and drug action results.

[0134] By jointly analyzing gene variation information and drug response information, combining gene variation with drug response can help doctors choose drugs that are more likely to be effective in patients and avoid using drugs that may cause serious adverse reactions. It can also predict adverse reactions, take preventive measures in advance, and reduce patient risks, thereby reducing medical costs and improving treatment efficiency.

[0135] In some optional implementations of this embodiment, the fusion output layer includes a feature fusion sublayer and a health prediction sublayer, and the fusion output submodule is further used for:

[0136] The feature fusion sublayer uses multidimensional data fusion technology to fuse the disease analysis results and the gene variation and drug action results to obtain fusion features;

[0137] Through the health prediction sublayer, individual feature recognition and health analysis are performed on the fusion features to obtain health analysis results.

[0138] By integrating the results of disease analysis with genetic variation and drug action results, it is possible to comprehensively analyze information from different data sources and dimensions, provide a more comprehensive health status assessment, improve the accuracy of diagnosis, help develop personalized treatment plans, and better manage patients' health conditions; it is able to predict possible diseases or health problems in the future, thereby achieving early intervention, which can reduce the need for expensive treatments and long-term care, thereby reducing overall medical costs.

[0139] In some optional implementations, the drug recommendation model includes a drug screening layer, a drug effect prediction layer, a recommendation result generation layer, and an explanation layer. The drug recommendation module 403 includes:

[0140] A drug screening submodule, configured to extract key information and perform semantic vector representation on the health analysis results, the standard medical record data and the standard gene data through the drug screening layer to obtain drug semantic features, and obtain candidate drugs through the drug semantic features;

[0141] A drug effect prediction submodule, for extracting key drug effect information from the health analysis results, the standard medical record data and the standard gene data through the drug effect prediction layer in combination with the candidate drug, to obtain drug efficacy and adverse reaction results;

[0142] A scoring submodule, for scoring the candidate drugs by combining the drug efficacy and adverse reaction results through the recommendation result generation layer, and screening out the optimal drug combination;

[0143] The annotation submodule is used to annotate the optimal drug combination with drug usage instructions through the explanation layer to obtain a drug recommendation plan.

[0144] By generating drug recommendation plans through drug recommendation models, personalized treatment can be provided based on the patient's specific condition, genotype, past treatment response and other personalized factors, making drug treatment more targeted and effective; by accurately matching drugs and patients, the treatment effect can be improved, ineffective treatment and drug waste can be reduced, thereby reducing medical costs; at the same time, the occurrence of adverse reactions can be reduced, and the quality and efficiency of overall medical services can be improved.

[0145] In some optional implementations of this embodiment, the drug effect prediction submodule is further used to:

[0146] In the drug effect prediction layer, a condition semantic vector is obtained by extracting key features from the health analysis results, the standard medical record data and the standard gene data;

[0147] Compare the standard gene data with the gene database to obtain SNP information of key gene sites, and obtain a drug effect feature vector based on the SNP information and the candidate drug;

[0148] The disease semantic vector and the drug effect feature vector are feature fused to obtain drug efficacy and adverse reaction results.

[0149] By predicting drug efficacy and adverse reactions, we can provide patients with more personalized treatment plans, improve drug treatment effects and medication safety, and reduce the risk of adverse reactions.

[0150] In some optional implementations, the drug recommendation device 400 further includes an updating module, which is used to:

[0151] Collecting feedback data from the target patient after using the recommended drug regimen, wherein the feedback data includes drug efficacy and adverse drug reactions;

[0152] The health analysis model and the drug recommendation model are updated according to the feedback data.

[0153] Based on feedback data, the health analysis model and drug recommendation model are retrained and optimized to continuously improve the accuracy and personalization of recommendations.

[0154] To solve the above technical problems, the present application also provides a computer device. Figure 5 , Figure 5 This is a basic structural block diagram of the computer device in this embodiment.

[0155] The computer device 5 includes a memory 51, a processor 52, and a network interface 53 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 5 with a memory 51, a processor 52, and a network interface 53, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0156] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0157] The memory 51 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 51 can be an internal storage unit of the computer device 5, such as a hard disk or memory of the computer device 5. In other embodiments, the memory 51 can also be an external storage device of the computer device 5, such as a plug-in hard disk equipped on the computer device 5, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. Of course, the memory 51 can also include both the internal storage unit of the computer device 5 and its external storage device. In this embodiment, the memory 51 is generally used to store the operating system and various application software installed on the computer device 5, such as computer-readable instructions of the drug recommendation method, etc. In addition, the memory 51 can also be used to temporarily store various types of data that have been output or are to be output.

[0158] The processor 52 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 52 is generally used to control the overall operation of the computer device 5. In this embodiment, the processor 52 is used to run computer-readable instructions or process data stored in the memory 51, such as computer-readable instructions for running the drug recommendation method.

[0159] The network interface 53 may include a wireless network interface or a wired network interface. The network interface 53 is generally used to establish a communication connection between the computer device 5 and other electronic devices.

[0160] By preprocessing medical record data and genetic data according to preset data standards and specifications, standard medical record data and standard genetic data can be obtained, which can reduce data noise and improve data quality. At the same time, data consistency can speed up calculations and improve the interpretability of data analysis. By fusing standard medical record data and standard genetic data for prediction through a health analysis model, multi-dimensional data in different feature spaces can be extracted to obtain comprehensive features of the target patients. At the same time, data from different sources can be fused to eliminate redundancy and noise, thereby improving the accuracy and reliability of the analysis. By analyzing health analysis results, standard medical record data and standard genetic data through a drug recommendation model, the level of personalization of drug recommendations can be significantly improved, making drug treatment more targeted and effective, and improving the safety of drug use for patients.

[0161] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the drug recommendation method as described above.

[0162] By preprocessing medical record data and genetic data according to preset data standards and specifications, standard medical record data and standard genetic data can be obtained, which can reduce data noise and improve data quality. At the same time, data consistency can speed up calculations and improve the interpretability of data analysis. By fusing standard medical record data and standard genetic data for prediction through a health analysis model, multi-dimensional data in different feature spaces can be extracted to obtain comprehensive features of the target patients. At the same time, data from different sources can be fused to eliminate redundancy and noise, thereby improving the accuracy and reliability of the analysis. By analyzing health analysis results, standard medical record data and standard genetic data through a drug recommendation model, the level of personalization of drug recommendations can be significantly improved, making drug treatment more targeted and effective, and improving the safety of drug use for patients.

[0163] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0164] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A drug recommendation method, characterized in that: The steps include: Obtaining medical record data and gene data of a target patient, and preprocessing the medical record data and gene data according to preset data standard specifications to obtain standard medical record data and standard gene data; Inputting the standard medical record data and the standard gene data into the trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain a health analysis result; The health analysis results, the standard medical record data and the standard gene data are input into the trained drug recommendation model to perform drug screening and drug effect prediction, and a drug recommendation plan is output.

2. The drug recommendation method according to claim 1, characterized in that: The health analysis model includes a feature extraction layer, a medical record analysis layer, a gene analysis layer and a fusion output layer. The step of inputting the standard medical record data and the standard gene data into the trained health analysis model and performing prediction by fusing the standard medical record data and the standard gene data to obtain the health analysis result includes: Inputting the standard medical record data and the standard gene data into the feature extraction layer to perform feature extraction, and obtaining corresponding medical record data features and gene data features; The medical record analysis layer fuses the medical record data features and the gene data features to perform analysis and prediction to obtain a disease analysis result; The gene data features are analyzed and predicted by the gene analysis layer to obtain gene variation and drug action results; The disease analysis results and the gene variation and drug action results are input into the fusion output layer, and the health analysis results are obtained through a data fusion algorithm.

3. The drug recommendation method according to claim 2, characterized in that: The gene analysis layer includes a gene identification sublayer and a drug effect analysis prediction sublayer. The step of analyzing and predicting the gene data features through the gene analysis layer to obtain gene variation and drug effect results includes: Comparing the gene sequences in the gene data features and identifying the variation sites through the gene identification sublayer to obtain gene variation information; According to the gene variation information, obtaining drug response information corresponding to the gene variation information by querying a pre-constructed gene variation and drug response association database; The gene variation information and the drug response information are input into the drug efficacy analysis prediction sublayer, and the gene variation information and the drug response information are jointly analyzed through a pre-trained large language model to obtain the gene variation and drug action results.

4. The drug recommendation method according to claim 2, characterized in that: The fusion output layer includes a feature fusion sublayer and a health prediction sublayer. The step of inputting the disease analysis result and the gene variation and drug action result into the fusion output layer and obtaining the health analysis result through the data fusion algorithm includes: The feature fusion sublayer uses multidimensional data fusion technology to fuse the disease analysis results and the gene variation and drug action results to obtain fusion features; Through the health prediction sublayer, individual feature recognition and health analysis are performed on the fusion features to obtain health analysis results.

5. The drug recommendation method according to claim 1, characterized in that: The drug recommendation model includes a drug screening layer, a drug effect prediction layer, a recommendation result generation layer and an explanation layer. The steps of inputting the health analysis results, the standard medical record data and the standard gene data into the trained drug recommendation model for drug screening and drug effect prediction and outputting a drug recommendation plan include: The drug screening layer extracts key information and represents semantic vectors from the health analysis results, the standard medical record data, and the standard gene data to obtain drug semantic features, and obtains candidate drugs through the drug semantic features; By combining the candidate drugs with the drug effect prediction layer, the key information of drug effects is extracted from the health analysis results, the standard medical record data and the standard gene data to obtain drug efficacy and adverse reaction results; The recommendation result generation layer combines the drug efficacy and adverse reaction results to score the candidate drugs and screen out the optimal drug combination; The optimal drug combination is annotated with drug usage instructions through the explanation layer to obtain a drug recommendation plan.

6. The drug recommendation method according to claim 5, characterized in that: The step of extracting key information of drug effects from the health analysis results, the standard medical record data and the standard gene data by combining the drug effect prediction layer with the candidate drug to obtain drug efficacy and adverse reaction results includes: In the drug effect prediction layer, a condition semantic vector is obtained by extracting key features from the health analysis results, the standard medical record data and the standard gene data; Compare the standard gene data with the gene database to obtain SNP information of key gene sites, and obtain a drug effect feature vector based on the SNP information and the candidate drug; The disease semantic vector and the drug effect feature vector are feature fused to obtain drug efficacy and adverse reaction results.

7. The drug recommendation method according to claim 1, characterized in that: After the step of outputting the drug recommendation scheme, the method further includes: Collecting feedback data from the target patient after using the recommended drug regimen, wherein the feedback data includes drug efficacy and adverse drug reactions; The health analysis model and the drug recommendation model are updated according to the feedback data.

8. A drug recommendation device, characterized in that: include: An acquisition module is used to acquire the medical record data and gene data of a target patient, and pre-process the medical record data and gene data according to a preset data standard specification to obtain standard medical record data and standard gene data; A health analysis module, used for inputting the standard medical record data and the standard gene data into a trained health analysis model, and performing prediction by fusing the standard medical record data and the standard gene data to obtain a health analysis result; The drug recommendation module is used to input the health analysis results, the standard medical record data and the standard gene data into the trained drug recommendation model to perform drug screening and drug effect prediction, and output a drug recommendation plan.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the drug recommendation method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the drug recommendation method according to any one of claims 1 to 7 are implemented.

Images