Intelligent intervention method for dyslipidemia based on multi-omics analysis

Abstract

The invention provides an intelligent intervention method for dyslipidemia based on multi-omics analysis, and aims to provide scientific and personalized treatment and lifestyle guidance for dyslipidemia patients. And the predictive ability of the deep learning model is combined with the interpretability and security verification mechanism of the knowledge graph and the rule engine, so that the scientificity, interpretability and security of each intervention scheme are ensured. An intervention scheme is generated through a deep learning model, and meanwhile, a knowledge graph and a rule engine are introduced, so that the interpretability, the safety and the dynamic adaptability of the scheme are ensured. By dynamically analyzing individualized genes, proteins and clinical characteristics, individualized treatment and lifestyle intervention suggestions are provided for patients. The model not only can significantly improve the curative effect and reduce the risk of adverse reactions, but also can realize continuous improvement of an individualized intervention scheme through real-time feedback and dynamic optimization, and provides an efficient blood fat management tool for doctors and patients.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for intelligent intervention of dyslipidemia based on multi-omics analysis, an electronic device and a computer-readable storage medium. Background Art

[0002] Hyperlipidemia is one of the important risk factors for atherosclerotic cardiovascular disease (ASCVD) and related complications.

[0003] At present, the management of hyperlipidemia mainly relies on drug interventions such as statins and PCSK9 inhibitors, but the efficacy of different individuals varies greatly. Therefore, the same type of hyperlipidemia medication and other treatment plans cannot be implemented for the same type of patients. Because there are large differences between individual patients with hyperlipidemia, some patients have different side effects or intolerance problems with traditional drug treatments due to differences in gene and protein expression levels. Therefore, clinical intervention for dyslipidemia. It is also necessary to adjust according to individual differences in patients, and traditional intervention methods for dyslipidemia are somewhat lacking. Diagnosis and treatment are based on the same type of abnormal blood dry blood conclusions, which lacks consideration of individual differences.

[0004] With the development of multi-omics technology, it is possible to combine genomics, proteomics, clinical biomarkers and lifestyle data to provide personalized intervention plans for each patient. However, there are currently few clinical studies to implement the above multi-omics intelligent intervention plans, and there is a lack of efficient and intelligent dyslipidemia and individualized intervention diagnosis and treatment capabilities. Summary of the invention

[0005] In order to solve the technical problems existing in the prior art, the present invention provides the following technical solutions: In one aspect, a method for intelligent intervention of dyslipidemia based on multi-omics analysis is provided, the method being implemented by an electronic device, the method comprising: S1. Collect multi-omics lipid data of patients with hyperlipidemia, including genomic data, proteomic data, clinical biomarkers and lifestyle data related to lipid indicators, as well as intervention measures; S2, inputting the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, and identifying and outputting a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; S3, inputting the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieves and outputs a personalized blood lipid intervention strategy that matches the multimodal feature vector; S4, writing the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; S5. Recommend the blood lipid diagnosis and treatment report to the patient and medical staff.

[0006] Preferably, the method for generating the multi-omics blood lipid intervention model comprises: Collect and preprocess the multi-omics lipid history big data of several hyperlipidemia patients, including: 1) Genomics data: Collect genetic polymorphism information related to lipid metabolism, drug metabolism and drug tolerance, including: • PCSK9, LDLR, APOB: related to cholesterol metabolism and the efficacy of PCSK9 inhibitors; • SLCO1B1: a gene that affects statin metabolism; • CETP, LIPC, APOA5: related to high-density lipoprotein (HDL) and low-density lipoprotein (LDL) metabolism; 2) Proteomics data: Mass spectrometry or protein chip technology is used to detect the expression level of specific proteins in blood samples; the collected indicators include: • PCSK9 protein concentration; • Apolipoproteins such as ApoB; • Inflammation-related proteins (C-reactive protein, CRP), which reflect the patient’s inflammatory status; 3) Clinical biomarkers: including the patient's lipid profile (total cholesterol, LDL, HDL and triglycerides), body mass index (BMI), blood pressure and glucose metabolism status; 4) Lifestyle data: A combination of questionnaire survey and wearable devices is used to collect dietary and running data of patients with hyperlipidemia; 5) Intervention data: Data on drug treatment and lifestyle adjustment interventions were collected through patient follow-up records and electronic health records (EHR); Performing feature extraction on the multi-omics blood lipid historical big data, extracting multi-omics blood lipid intervention features from the multi-omics blood lipid historical big data of each hyperlipidemia patient, including genomic features, proteomic features, clinical biomarkers, lifestyle data features, and intervention measures features; Annotating the multi-omics blood lipid intervention characteristics, wherein the annotation information is the intervention effect level determined according to the multi-omics blood lipid intervention characteristics; Statistically annotating the multi-omics blood lipid intervention features of each hyperlipidemia patient and constructing a feature set; Dividing the feature set into a training set and a validation set according to a preset ratio; Inputting the training set into a preset machine learning model to perform feature training and learning to generate the multi-omics blood lipid intervention model; The application performance of the multi-omics blood lipid intervention model was verified using the validation set: If the verification is passed, the multi-omics blood lipid intervention model is deployed on the background server for intervention services for patients with hyperlipidemia; If the verification fails, the above steps are repeated to reconstruct the multi-omics blood lipid intervention model.

[0007] Preferably, when performing feature training and learning, it includes: Initial training: performing initial training on the training set by supervised learning training to generate the initial multi-omics blood lipid intervention model; Reinforced training: Based on Q-Learning or deep reinforcement learning algorithm, the multi-omics lipid intervention model is subjected to reinforced training and learning according to clinical feedback on the model output.

[0008] Preferably, the method for constructing the blood lipid intervention knowledge graph and rule engine includes: Based on the genomic characteristics, proteomic characteristics and corresponding medications of patients with hyperlipidemia, a mapping relationship between genes, proteins and drugs is constructed, and a corresponding blood lipid intervention knowledge graph is constructed based on the mapping relationship; Configuring a rule engine for the blood lipid intervention knowledge graph includes: First-tier rules: corresponding to different drugs based on and / or protein expression levels; The second level of rules: doses or courses corresponding to different ApoB concentrations and CRP levels; After the configuration is completed, the blood lipid intervention knowledge graph and rule engine are obtained.

[0009] On the other hand, a device for intelligent intervention of dyslipidemia based on multi-omics analysis is provided, and the device for intelligent intervention of dyslipidemia based on multi-omics analysis is used to implement the above-mentioned method for intelligent intervention of dyslipidemia based on multi-omics analysis, and the device comprises: The data collection end is used to collect multi-omics blood lipid data of patients with hyperlipidemia and upload it to the backend server; A backend server is used to input the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, identify and output a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; and input the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieve and output a personalized blood lipid intervention strategy that matches the multimodal feature vector; and write the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; and recommend the blood lipid diagnosis and treatment report to the patient end and the medical care end; The patient end is used to receive and view the blood lipid diagnosis and treatment report; A medical terminal, for receiving and viewing the blood lipid diagnosis and treatment report; The data collection end, the patient end and the medical end are respectively connected to the backend server for communication.

[0010] On the other hand, an electronic device is provided, comprising: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned intelligent intervention methods for dyslipidemia based on multi-omics analysis is implemented.

[0011] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned intelligent intervention methods for dyslipidemia based on multi-omics analysis.

[0012] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: This paper proposes a multimodal, deep learning, knowledge graph and rule engine-based personalized intervention program for dyslipidemia, which aims to provide scientific, personalized treatment and lifestyle guidance for patients with dyslipidemia through the deep integration of multi-omics features, intervention program data and lifestyle information. The predictive ability of the deep learning model is combined with the interpretability and safety verification mechanism of the knowledge graph and rule engine to ensure the scientificity, interpretability and safety of each intervention program.

[0013] The present invention generates intervention plans through deep learning models, and introduces knowledge graphs and rule engines to ensure the interpretability, safety, and dynamic adaptability of the plans. By dynamically analyzing individualized genes, proteins, and clinical characteristics, personalized treatment and lifestyle intervention recommendations are provided to patients. The model can not only significantly improve the efficacy and reduce the risk of adverse reactions, but also achieve continuous improvement of individualized intervention plans through real-time feedback and dynamic optimization, providing doctors and patients with efficient blood lipid management tools. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 This is a flow chart of an intelligent intervention method for dyslipidemia based on multi-omics analysis provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of a model training process provided by an embodiment of the present invention; Figure 3 It is a block diagram of an intelligent intervention device for dyslipidemia based on multi-omics analysis provided by an embodiment of the present invention; Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0017] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0018] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0019] In the embodiments of the present invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are consistent.

[0020] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0021] The embodiment of the present invention provides an intelligent intervention method for dyslipidemia based on multi-omics analysis, which can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The flowchart of the intelligent intervention method for dyslipidemia based on multi-omics analysis is shown, and the processing flow of the method may include the following steps: S1. Collect multi-omics lipid data of patients with hyperlipidemia, including genomic data, proteomic data, clinical biomarkers and lifestyle data related to lipid indicators, as well as intervention measures; S2, inputting the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, and identifying and outputting a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; S3, inputting the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieves and outputs a personalized blood lipid intervention strategy that matches the multimodal feature vector; S4, writing the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; S5. Recommend the blood lipid diagnosis and treatment report to the patient and medical staff.

[0022] The present invention uses a deep learning model to generate an intervention plan for hyperlipidemia patients, and introduces a knowledge graph and a rule engine to ensure the interpretability, safety, and dynamic adaptability of the plan. By dynamically analyzing individualized genes, proteins, and clinical characteristics, personalized treatment and lifestyle intervention recommendations are provided to patients. The model can not only significantly improve the efficacy and reduce the risk of adverse reactions, but also achieve continuous improvement of individualized intervention plans through real-time feedback and dynamic optimization, providing doctors and patients with efficient blood lipid management tools.

[0023] The main technical logic and applications are as follows: 1. Deep learning model outputs preliminary intervention plan • The deep learning model generates personalized intervention plans based on multi-omics data and patient characteristics, including drug selection, dosage, and lifestyle adjustment recommendations.

[0024] • Given sufficient data, the model can capture complex multi-dimensional data relationships, thereby providing refined personalized recommendations.

[0025] 2. Supplementary support for knowledge graphs • The knowledge graph contains professional medical knowledge on dyslipidemia and drug-gene interactions, which can supplement the model's recommendations in areas where data is scarce.

[0026] • Especially in scenarios such as rare diseases, special populations, or new drugs, knowledge graphs provide additional clinical knowledge support for the model to ensure the applicability of the solution.

[0027] 3. Constraint verification of rule engine • The rule engine verifies the safety of the model output according to clinical guidelines and drug safety standards, for example, avoiding the use of specific drugs for patients with certain genotypes, or reminding patients of interaction contraindications.

[0028] • During the solution generation process, the rule engine can detect and correct the model output in real time according to the preset security rules to enhance the security of the solution.

[0029] 4. Dynamic adaptability and personalized optimization • The knowledge graph and rule engine work together with the model output to enable the system to integrate the latest medical research advances. For example, by updating the knowledge graph, the system can automatically match the appropriate population for new therapies or new drugs.

[0030] • The reinforcement learning mechanism enables the model to continuously absorb patient feedback data and dynamically adjust intervention recommendations to make the plan more tailored to individual needs.

[0031] The following will mainly describe the application principles of the present invention, especially the AI ​​model, knowledge graph and rule engine.

[0032] like Figure 2 The generation process of the multi-omics lipid intervention model shown in the figure mainly includes the following contents: 1. Data collection and preprocessing 1) Genomic data: Collect saliva or blood samples and use next-generation sequencing technology (such as Illumina, PacBio) to detect gene polymorphisms. Collect gene polymorphism information related to lipid metabolism, drug metabolism and drug tolerance, including: • PCSK9, LDLR, APOB: related to cholesterol metabolism and the efficacy of PCSK9 inhibitors; • SLCO1B1: a gene that affects statin metabolism; • CETP, LIPC, APOA5: related to high-density lipoprotein (HDL) and low-density lipoprotein (LDL) metabolism, which helps in personalized lipid metabolism assessment.

[0033] Proteomics data: Mass spectrometry or protein chip technology is used to detect the expression level of specific proteins in blood samples. The detection instruments include liquid chromatography-mass spectrometry system (LC-MS / MS) and quantitative protein chip.

[0034] 2) Collection indicators include: • PCSK9 protein concentration; • Apolipoproteins such as ApoB; • Inflammation-related proteins (C-reactive protein, CRP) reflect the patient’s inflammatory status.

[0035] 3) Clinical biomarkers: including the patient's blood lipid indicators (total cholesterol, LDL, HDL, triglycerides), body mass index (BMI), blood pressure, and glucose metabolism status.

[0036] 4) Lifestyle data: collected through a combination of questionnaires and wearable devices, as follows: • Questionnaires: Use digital health questionnaires (such as apps or web forms) to obtain self-reported data on diet, exercise, smoking, and drinking.

[0037] • Wearable devices: such as smart bracelets or smart watches that automatically record exercise frequency, intensity, number of steps, etc. Dietary information can be uploaded by taking photos with a mobile phone or recorded by scanning a code.

[0038] 5) Intervention data: This includes both drug treatment and lifestyle adjustments. Intervention data are collected through patient follow-up records, electronic health records (EHR), and other sources.

[0039] Data standardization Clean and standardize genetic, protein, clinical, and lifestyle data to ensure uniformity and comparability of each data type.

[0040] 2. Feature extraction 1) Genomic characteristics Through the following feature engineering processing: • Gene variation coding: Gene loci are encoded using a “0 / 1 / 2” coding pattern (e.g., 0 for wild type, 1 for heterozygous mutation, and 2 for homozygous mutation).

[0041] • Polymorphism feature interaction: Through multivariate interaction feature analysis, the SNP (single nucleotide polymorphism) information significantly associated with lipid metabolism is fused.

[0042] 2) Proteomic characteristics • Standardization: Normalize or Z-score the protein concentration data to eliminate the numerical deviation between different samples.

[0043] • Dimensionality reduction: Use methods such as principal component analysis (PCA) to extract protein features that are highly correlated with lipid metabolism to reduce data dimensions and computational complexity.

[0044] 3) Clinical biomarkers • Interval division: For example, blood lipid levels are divided into normal, slightly abnormal and highly abnormal.

[0045] • Interactive features: Combine BMI, blood sugar, blood pressure and other data to construct interactive features for assessing cardiovascular health risks.

[0046] 4) Lifestyle data characteristics • Diet: Diet healthiness is quantified using dietary scoring criteria (e.g., MedDiet Score), coded as a continuous variable.

[0047] • Exercise: Quantify the impact of exercise using interactive features of exercise frequency, intensity, and type.

[0048] • Smoking / drinking: The frequency of smoking and drinking is categorized and coded (e.g., light, moderate, heavy) to facilitate model analysis.

[0049] 5) Intervention feature extraction Intervention frequency and intensity characteristics • Medication: Numerical coding of medication type, dosage, and frequency. For example, dosage is expressed in mg / dL, and frequency is expressed in daily or weekly doses.

[0050] • Lifestyle interventions: such as METs and weekly frequency of exercise to capture the impact of lifestyle changes on blood lipids.

[0051] Time dimension features • Use the sliding window method to calculate the change rate of the intervention effect at different time points. For example, the intervention effect can be reflected by the change rate of blood lipids (Δ value) in the most recent month, which is used as a feature input model.

[0052] • Construct short-term (one month) and long-term (more than half a year) window features to capture trends in intervention effects over time.

[0053] Intervention combination characteristics • Record the combined effects of medication and lifestyle interventions and construct interaction features (e.g., the combination of “high-intensity exercise + low-fat diet”).

[0054] • For multiple interventions, use independent features to characterize the additive effects of different combinations.

[0055] Feedback Features • Construct compliance and side effect scoring features through patient feedback data. For example, grading compliance and side effects can help the model determine the suitability of different intervention plans.

[0056] • Quality of life scores were recorded to assess the impact of the intervention on overall health status.

[0057] Individual Difference Characteristics • Use the patient’s basic genetic characteristics (such as PCSK9 genotype) to interactively analyze intervention data to extract personalized intervention effects.

[0058] • Combine protein features (e.g., inflammation-related protein levels) to construct interactive signatures for determining the adaptability of personalized interventions.

[0059] 4. Multimodal Deep Learning Model Model design: The model adopts a multimodal deep learning structure, integrates genomics, proteomics, clinical biomarkers, and lifestyle data as input, and is trained through a reinforcement learning framework.

[0060] Model training: 1) Initial training: The collected data set is divided into a training set and a validation set, and the initial multimodal deep learning model (such as Transformer or multimodal neural network) is trained. The initial model uses supervised learning to establish a mapping relationship between input features and intervention plans using the data of known hyperlipidemia patients and intervention effects as labels, and learns the relationship between multimodal input and intervention effects.

[0061] The initial model training can be specifically combined with the structure and principle of the model used. For example, the model training steps using Transformer are generally as follows: 1). Data preparation ‌Training set‌: Prepare a training set containing a large number of labeled samples. These datasets usually contain various language phenomena, structures, and patterns. In the field of natural language processing (NLP), commonly used datasets include translation datasets from the WMT (Workshop on Machine Translation) series.

[0062] Validation set: Prepare a validation set for model performance verification. The dataset should be independent of the training set but come from the same distribution.

[0063] It can be divided into 8:2.

[0064] 2) Data preprocessing ‌Text cleaning‌: remove HTML tags, special characters, URL links and other information not related to the task.

[0065] Tokenization: Split text into smaller units (such as words or subwords) and convert these units into tokens (such as integer IDs) that the model can process.

[0066] ‌Build Vocabulary‌: Extract all unique tokens from the training data and build a vocabulary.

[0067] ‌Numerization‌: Convert text data into numerical form for model processing.

[0068] ‌Add special markers‌: Add special markers at the beginning and end of the sequence, such as <s> and< / s> , to indicate sentence boundaries.

[0069] 3). Model construction and training ‌Select Transformer Architecture‌: Select an appropriate Transformer model architecture, such as a basic model or a large model, based on task requirements.

[0070] ‌Set hyperparameters‌: Set hyperparameters such as learning rate, batch size, optimizer (such as Adam), etc.

[0071] ‌Load training data‌: Load the preprocessed training data into the model.

[0072] ‌Model training‌: The model is iteratively trained using the training data, and the loss function is calculated and the model weights are updated for each batch. At the end of each training cycle (epoch), the model performance is evaluated using the validation set, and key indicators (such as accuracy and loss value) are recorded.

[0073] ‌Save Model‌: Save model weights and parameters periodically during training so that you can resume training or perform model evaluation when needed.

[0074] 4). Model evaluation and optimization ‌Evaluate model performance using validation set‌: During the training process, use the validation set to evaluate the performance of the model to ensure that the model is not overfitting. Evaluation metrics can include accuracy, recall, F1 score, etc.

[0075] ‌Adjust hyperparameters‌: Based on the performance on the validation set, adjust hyperparameters such as learning rate and batch size to optimize model performance.

[0076] ‌Try different model architectures‌: If the current model architecture does not perform well, you can try a different model architecture or improve the existing architecture.

[0077] Regularization technology: Use regularization, dropout and other techniques to prevent overfitting and improve the generalization ability of the model.

[0078] 5). Model deployment and application ‌Model deployment‌: Deploy the trained model to the production environment (the backend server in this case) for practical application.

[0079] ‌Continuous Monitoring and Optimization‌: Continuously monitor model performance in actual applications and make further optimizations and adjustments as needed.

[0080] 2) Reinforcement learning training: Introduce reinforcement learning strategies (such as Q-Learning or deep reinforcement learning) and add feedback loops to the model. Adaptively optimize the model based on patient treatment feedback (such as efficacy and adverse reactions) to adjust the recommended plan and provide dynamic personalized treatment recommendations.

[0081] • Reward mechanism: The model simulates follow-up feedback from patients, using improved lipid levels and reduced adverse reactions as reward signals to continuously adjust the model’s recommendation strategy.

[0082] • Dynamic optimization: Based on real-time feedback data, the model updates parameters to improve the effectiveness of the recommended individualized intervention plan.

[0083] You can refer to the following strengthening steps: 1). Determine the reinforcement learning framework ‌Choose an algorithm‌: Decide whether to use Q-Learning or deep reinforcement learning (such as DQN, i.e. deep Q network). Q-Learning is suitable for situations where the state space and action space are relatively small, while deep reinforcement learning is more suitable for processing high-dimensional data.

[0084] ‌Define state space‌: The state space should include all factors that may affect the recommendation decision, such as the patient's physiological indicators, historical treatment records, genetic information, etc.

[0085] ‌Define the action space‌: The action space should include all possible recommendation options, such as drug types, dosages, dosing frequency, etc.

[0086] ‌Define reward function‌: The reward function should be defined based on the patient’s treatment feedback. For example, if the treatment is effective and has fewer adverse reactions, a positive reward is given; otherwise, a negative reward is given.

[0087] 2). Data collection and preprocessing Collect patient data: including the patient’s physiological indicators, medical history, genetic information, etc.

[0088] ‌Collect treatment feedback‌: including patient efficacy evaluation, adverse reaction reports, etc.

[0089] ‌Data preprocessing‌: Cleaning, transforming, and standardizing the data to ensure that the algorithm can process it correctly.

[0090] 3). Model training Initialize model parameters: including learning rate, discount factor, exploration rate, etc.

[0091] ‌Iterative Training‌: Iterative training is performed using patient data and treatment feedback. At each time step, the model selects an action (i.e., a recommendation) based on the current state, and then updates the Q value or neural network parameters based on the treatment feedback.

[0092] ‌Exploration and Utilization‌: In the early stages of training, the model should be more inclined to explore new recommendation schemes to discover better strategies; in the later stages of training, it should make more use of the learned knowledge to make recommendations.

[0093] 4). Model evaluation and optimization ‌Evaluate model performance using a validation set‌: Regularly evaluate model performance using a validation set during training to ensure that the model is not overfitting.

[0094] ‌Adjust hyperparameters‌: Adjust hyperparameters such as learning rate and discount factor based on the performance on the validation set.

[0095] Optimize reward function: Adjust the reward function according to actual needs to better reflect the treatment goals.

[0096] By introducing reinforcement learning strategies and adding feedback loops to the model, the model can be adaptively optimized based on patient treatment feedback to adjust the recommended plan, thereby improving treatment outcomes and patient satisfaction.

[0097] Preferably, the method for constructing the blood lipid intervention knowledge graph and rule engine comprises: Based on the genomic characteristics, proteomic characteristics and corresponding medications of patients with hyperlipidemia, a mapping relationship between genes, proteins and drugs is constructed, and a corresponding blood lipid intervention knowledge graph is constructed based on the mapping relationship; Configuring a rule engine for the blood lipid intervention knowledge graph includes: First-tier rules: corresponding to different drugs based on and / or protein expression levels; The second level of rules: doses or courses corresponding to different ApoB concentrations and CRP levels; After the configuration is completed, the blood lipid intervention knowledge graph and rule engine are obtained.

[0098] The knowledge graph includes various graph nodes (various features and recommended strategies (drugs) in this embodiment) and the relationships between the nodes.

[0099] Through the above-mentioned feature engineering, we can find the medication relationship between genomic features, proteomic features and corresponding medications, thereby constructing a knowledge graph and configuring corresponding dosage and treatment rules, so as to configure corresponding dosage and treatment rules for the lipid medication strategy recommended by the graph.

[0100] Knowledge graph construction can be understood and implemented in combination with existing technologies.

[0101] Here, please refer to the following steps: 1. Data Collection and Preprocessing 1.1 Data Collection Collect genomic data (such as gene variation, gene expression, etc.) of patients with hyperlipidemia.

[0102] Collect corresponding proteomics data (such as protein expression levels, protein interactions, etc.).

[0103] Collect patients' medication records, including drug type, dosage, course of treatment, and treatment effect.

[0104] 1.2 Data Preprocessing Perform quality control on genomic and proteomics data to remove low-quality or abnormal data.

[0105] Clean and standardize medication records to ensure the accuracy of information such as drug names and dosages.

[0106] Integrate genomics, proteomics and medication data into a unified dataset.

[0107] 2. Constructing gene-protein-drug mapping relationships Bioinformatics methods (such as gene-protein association analysis, drug target prediction, etc.) are used to analyze genomic and proteomic data to identify genes and proteins related to hyperlipidemia.

[0108] According to the drug's mechanism of action and target information, the drug is associated with related genes and proteins to construct a mapping relationship between gene-protein-drug.

[0109] 3. Constructing a knowledge graph for blood lipid intervention Based on the mapping relationship between gene, protein and drug, a knowledge graph of blood lipid intervention is constructed. The nodes of the graph can include genes, proteins, drugs, etc., and the edges represent the associations between them.

[0110] Add attributes to each node and edge in the graph, such as gene mutation type, protein expression level, drug dosage and course of treatment, etc.

[0111] The blood lipid intervention knowledge graph is visualized using the graph visualization tool to facilitate subsequent analysis and application.

[0112] 4. Configure the rule engine 4.1 First layer rule configuration Based on genomic and proteomic data, a list of recommended drugs is set for different gene variants or protein expression levels.

[0113] When a patient's genomic or proteomic characteristics meet specific conditions, the rule engine can automatically recommend corresponding drugs.

[0114] 4.2 Layer 2 rule configuration Recommended rules for different doses or courses of treatment are set based on the patient's ApoB concentration and CRP level.

[0115] When a patient's ApoB concentration or CRP level reaches a certain threshold, the rules engine can automatically adjust the dosage or duration of the drug.

[0116] 5. Integration and testing Integrate the lipid intervention knowledge graph with the rule engine to ensure that the two can work together.

[0117] The integrated system is tested to verify its recommendation effect and accuracy in different scenarios.

[0118] Optimize and adjust the system based on the test results to improve its practicality and reliability.

[0119] 6. Deployment and Application The integrated blood lipid intervention knowledge graph and rule engine are deployed to the background for use by doctors or medical workers.

[0120] Through the system interface or API interface, doctors can input patients' genomic, proteomic and clinical data to obtain personalized lipid intervention recommendations.

[0121] The system is updated and maintained regularly to ensure it keeps pace with the latest medical research results and clinical practices.

[0122] Through the above steps, we can successfully build a blood lipid intervention knowledge graph based on the genomic characteristics, proteomic characteristics and medications of patients with hyperlipidemia, and configure the corresponding rule engine to provide personalized treatment plans for patients with hyperlipidemia.

[0123] Combination Figure 3 As shown in the figure, an AI model for identifying clinical features of blood lipids (multi-omics blood lipid intervention model) and a knowledge graph and rule engine for recommending blood lipid treatment strategies are deployed in the background to reflect the following technical values: 1. Personalized drug intervention: Using genomic and proteomic data, more accurate recommendations on drug type, dosage, and treatment frequency can be made; 2. Dynamically optimize treatment plans: Based on the patient's biomarker feedback, the model can dynamically adjust the recommended results to achieve personalized optimization; 3. Improve efficacy and safety: Personalized treatment plans provided by multi-omics analysis models are expected to improve treatment outcomes and reduce adverse drug reactions.

[0124] Figure 3 is a block diagram of a device for intelligent intervention of dyslipidemia based on multi-omics analysis according to an exemplary embodiment, and the device is used for an intelligent intervention method of dyslipidemia based on multi-omics analysis. Figure 3 , the device comprises: The data collection end is used to collect multi-omics blood lipid data of patients with hyperlipidemia and upload it to the backend server; A backend server is used to input the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, identify and output a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; and input the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieve and output a personalized blood lipid intervention strategy that matches the multimodal feature vector; and write the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; and recommend the blood lipid diagnosis and treatment report to the patient end and the medical care end; The patient end is used to receive and view the blood lipid diagnosis and treatment report; A medical terminal, for receiving and viewing the blood lipid diagnosis and treatment report; The data collection end, the patient end and the medical end are respectively connected to the backend server for communication.

[0125] For data sharing and storage between the background and various types of data collection terminals, please refer to the data collection process of hospital talent, testing and other ports.

[0126] The backend can manage the backend communication with the patient side and the medical side through APP, or mini-programs, etc. It can be implemented and understood in combination with the existing hospital terminals, mini-programs or web pages for managing doctors and patients.

[0127] Please understand the interaction between the above background and each terminal in conjunction with the corresponding steps in the above method, and will not be repeated here.

[0128] Figure 4 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 4 As shown, the electronic device 410 may include a first processor 2001 .

[0129] Optionally, the electronic device 410 may further include a memory 2002 and a transceiver 2003 .

[0130] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0131] Combine the following Figure 4 The components of the electronic device 410 are described in detail: The first processor 2001 is the control center of the electronic device 410, and may be a processor or a general term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or may be application specific integrated circuits (ASICs), or may be one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (field programmable gate arrays, FPGAs).

[0132] Optionally, the first processor 2001 can perform various functions of the electronic device 410 by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002 .

[0133] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 CPU0 and CPU1 are shown in FIG.

[0134] In a specific implementation, as an embodiment, the electronic device 410 may also include multiple processors, such as Figure 4 The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0135] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled to be executed by the first processor 2001. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0136] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001, or may exist independently and access the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0137] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0138] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 4 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0139] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently and communicate with the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0140] It should be noted that Figure 4 The structure of the electronic device 410 shown in the figure does not constitute a limitation on the router, and the actual knowledge structure recognition device may include more or fewer components than those shown in the figure, or combine certain components, or arrange the components differently.

[0141] In addition, the technical effects of the electronic device 410 can refer to the technical effects of the intelligent intervention method for dyslipidemia based on multi-omics analysis described in the above method embodiment, which will not be repeated here.

[0142] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0143] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0144] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0145] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0146] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0147] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0148] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0149] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0150] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0151] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0152] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0153] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0154] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An intelligent intervention method for dyslipidemia based on multi-omics analysis, characterized in that: The method comprises: S1. Collect multi-omics lipid data of patients with hyperlipidemia, including genomic data, proteomic data, clinical biomarkers and lifestyle data related to lipid indicators, as well as intervention measures; S2, inputting the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, and identifying and outputting a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; S3, inputting the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieves and outputs a personalized blood lipid intervention strategy that matches the multimodal feature vector; S4, writing the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; S5. Recommend the blood lipid diagnosis and treatment report to the patient and medical staff.

2. The intelligent intervention method for dyslipidemia based on multi-omics analysis according to claim 1, characterized in that: The method for generating the multi-omics blood lipid intervention model comprises: Collect and preprocess the multi-omics lipid history big data of several hyperlipidemia patients, including: 1) Genomics data: Collect genetic polymorphism information related to lipid metabolism, drug metabolism and drug tolerance, including: • PCSK9, LDLR, APOB: related to cholesterol metabolism and the efficacy of PCSK9 inhibitors; • SLCO1B1: a gene that affects statin metabolism; • CETP, LIPC, APOA5: related to high-density lipoprotein (HDL) and low-density lipoprotein (LDL) metabolism; 2) Proteomics data: Mass spectrometry or protein chip technology is used to detect the expression level of specific proteins in blood samples; the collected indicators include: • PCSK9 protein concentration; • Apolipoproteins such as ApoB; • Inflammation-related proteins (C-reactive protein, CRP), which reflect the patient’s inflammatory status; 3) Clinical biomarkers: including the patient's lipid profile (total cholesterol, LDL, HDL and triglycerides), body mass index (BMI), blood pressure and glucose metabolism status; 4) Lifestyle data: A combination of questionnaire survey and wearable devices is used to collect dietary and running data of patients with hyperlipidemia; 5) Intervention data: Data on drug treatment and lifestyle adjustment interventions were collected through patient follow-up records and electronic health records (EHR); Performing feature extraction on the multi-omics blood lipid historical big data, extracting multi-omics blood lipid intervention features from the multi-omics blood lipid historical big data of each hyperlipidemia patient, including genomic features, proteomic features, clinical biomarkers, lifestyle data features, and intervention measures features; Annotating the multi-omics blood lipid intervention characteristics, wherein the annotation information is the intervention effect level determined according to the multi-omics blood lipid intervention characteristics; Statistically annotating the multi-omics blood lipid intervention features of each hyperlipidemia patient and constructing a feature set; Dividing the feature set into a training set and a validation set according to a preset ratio; Inputting the training set into a preset machine learning model to perform feature training and learning to generate the multi-omics blood lipid intervention model; The application performance of the multi-omics blood lipid intervention model was verified using the validation set: If the verification is passed, the multi-omics blood lipid intervention model is deployed on the background server for intervention services for patients with hyperlipidemia; If the verification fails, the above steps are repeated to reconstruct the multi-omics blood lipid intervention model.

3. The intelligent intervention method for dyslipidemia based on multi-omics analysis according to claim 2, characterized in that: When performing feature training and learning, it includes: Initial training: performing initial training on the training set by supervised learning training to generate the initial multi-omics blood lipid intervention model; Reinforced training: Based on Q-Learning or deep reinforcement learning algorithm, the multi-omics lipid intervention model is subjected to reinforced training and learning according to clinical feedback on the model output.

4. The intelligent intervention method for dyslipidemia based on multi-omics analysis according to claim 2, characterized in that: The method for constructing the blood lipid intervention knowledge graph and rule engine includes: Based on the genomic characteristics, proteomic characteristics and corresponding medications of patients with hyperlipidemia, a mapping relationship between genes, proteins and drugs is constructed, and a corresponding blood lipid intervention knowledge graph is constructed based on the mapping relationship; Configuring a rule engine for the blood lipid intervention knowledge graph includes: First-tier rules: corresponding to different drugs based on and / or protein expression levels; The second level of rules: doses or courses corresponding to different ApoB concentrations and CRP levels; After the configuration is completed, the blood lipid intervention knowledge graph and rule engine are obtained.

5. An intelligent intervention device for dyslipidemia based on multi-omics analysis, wherein the intelligent intervention device for dyslipidemia based on multi-omics analysis is used to implement the intelligent intervention method for dyslipidemia based on multi-omics analysis as claimed in any one of claims 1 to 4, characterized in that: The device comprises: The data collection end is used to collect multi-omics blood lipid data of patients with hyperlipidemia and upload it to the backend server; A backend server is used to input the multi-omics blood lipid data into a preset multi-omics blood lipid intervention model, identify and output a multimodal feature vector of blood lipid indicators in the multi-omics blood lipid data through the multi-omics blood lipid intervention model; and input the multimodal feature vector into a preset blood lipid intervention knowledge graph and rule engine, and the blood lipid intervention knowledge graph and rule engine retrieve and output a personalized blood lipid intervention strategy that matches the multimodal feature vector; and write the personalized blood lipid intervention strategy and the gene and protein features in the multimodal feature vector into a preset diagnosis and treatment report to generate a corresponding blood lipid diagnosis and treatment report; and recommend the blood lipid diagnosis and treatment report to the patient end and the medical care end; The patient end is used to receive and view the blood lipid diagnosis and treatment report; A medical terminal, for receiving and viewing the blood lipid diagnosis and treatment report; The data collection end, the patient end and the medical end are respectively connected to the backend server for communication.

6. An electronic device, characterized in that: The electronic device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 4.

Images