AI evaluation-based atherosclerotic plaque reversal strategy recommendation method

Abstract

According to the atherosclerotic plaque reversal strategy recommendation method based on AI evaluation, comprehensive analysis of multi-modal data is realized through an AI large model algorithm, and a personalized intervention scheme based on dynamic risk grading is provided for doctors, so that acute cardiovascular events are reduced, and long-term prognosis of patients is improved. A set of intelligent risk grading system is constructed by using a large model algorithm and multi-modal data, and drug and nutritional supplement recommendation is performed through a large model, so that atherosclerotic plaque reversal of a patient is realized. Clinical artery image data of a patient can be processed, and risk assessment and grading are performed through the trained AI model to generate a personalized intervention scheme. Compared with a system based on rules, the AI algorithm has higher flexibility and robustness through large-scale data training, can dynamically predict the risk level of the patient according to the multi-dimensional features of the patient and provide an accurate intervention scheme, and meets the requirements of different patients.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to a method for recommending an atherosclerotic plaque reversal strategy based on AI evaluation, an atherosclerotic plaque reversal strategy recommendation device based on AI evaluation, an electronic device, and a computer-readable storage medium. Background Art

[0002] Atherosclerosis is the main cause of cardiovascular diseases such as coronary heart disease and stroke, and is characterized by the formation and deposition of atherosclerotic plaques on the inner wall of arterial blood vessels. The presence of plaques will gradually lead to blood flow restriction and significantly increase the risk of acute cardiovascular events. Existing atherosclerosis management mainly focuses on the control of blood lipid indicators, especially the reduction of low-density lipoprotein cholesterol (LDL-C) to delay the progression of plaques. However, for high-risk plaques that have already formed, traditional lipid-lowering treatments and follow-up evaluation methods have limited effects, especially in terms of individualized intervention and risk event management.

[0003] In recent years, the combination of coronary CT angiography (CCTA) and the international CAD-RADS 2.0 standard can identify high-risk plaque characteristics, and the development of artificial intelligence (AI) and large model algorithms has made individualized management of risk stratification possible. However, it still faces the following clinical problems: morphological and functional changes may occur during the follow-up of atherosclerotic plaques, and traditional rule-defined methods lack flexibility in dealing with the multidimensional characteristics of different patients and cannot achieve real-time individualized adjustments.

[0004] Therefore, it is necessary to provide a more optimized intelligent prediction solution to provide doctors with personalized intervention plans based on dynamic risk stratification, and recommend drugs and nutritional supplements through large models to achieve reversal of patients' atherosclerotic plaques, thereby reducing acute cardiovascular events and improving patients' long-term prognosis. 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: On the one hand, a method for recommending an atherosclerotic plaque reversal strategy based on AI evaluation is provided, the method being implemented by an electronic device, the method comprising: S1. Dataset construction: collect and preprocess a large-scale dataset related to atherosclerosis, including CTA images, FFR-CT values, LDL-C values, and clinical history data; S2. Multimodal data processing: using CNN to extract high-risk plaque features in the CTA image, using GNN to analyze the spatial distribution and structural relationship of the high-risk plaque features, and using the self-attention mechanism to capture the corresponding FFR-CT value, LDL-C value and clinical risk factors for the high-risk plaque features; S3. Risk grading generation: generating individualized risk grading features based on the characteristic relationship of the high-risk plaque features described above; S4. Personalized intervention plan generation: generating and binding corresponding individualized plaque reversal strategy treatment plans for different risk grading characteristics, wherein the plans include drug combinations and nutritional supplements corresponding to the risk grading characteristics; S5. Model training and application: Input the risk grading features into a preset machine learning model for model training to generate a corresponding AI recommendation model for plaque reversal strategy. After clinical verification, the AI ​​recommendation model for plaque reversal strategy is deployed in the background and put into use.

[0006] Preferably, in step S2, after extracting the high-risk plaque features in the CTA image using CNN, the step further includes: The high-risk plaque features are marked, and the marked objects include: Positive remodeling: the blood vessels dilate at the plaque, and the ratio of the outer vessel diameter to the average diameter of the proximal and distal ends is greater than 1.1; Low-attenuation plaque: non-calcified plaque with a density less than 30 HU, which is associated with the risk of rupture; Spotty calcification: small calcification spots within the plaque, indicating high activity; Napkin ring sign: an area of ​​low attenuation bordering a ring-like structure of higher CT attenuation, suggesting a lipid-core plaque with a thin fibrous cap.

[0007] Preferably, in step S2, the corresponding FFR-CT value, LDL-C value and clinical risk factors are captured for the high-risk plaque feature through a self-attention mechanism, including: Parsing the clinical history data, and capturing corresponding clinical risk factors for the labeled objects of the high-risk plaque features through a self-attention mechanism; At the same time, the corresponding FFR-CT value and LDL-C value are captured for the labeled object of the high-risk plaque feature through the self-attention mechanism.

[0008] Preferably, in step S5, the plaque reversal strategy AI recommendation model is deployed in the background and put into use, including: Inputting the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; Identify the high-risk plaque features in the patient's CTA images through the plaque reversal strategy AI recommendation model, combine the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data, and output the risk grading features of the corresponding patient; According to the pre-bound feature binding relationship, an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics is matched and output.

[0009] Preferably, step S4 further includes: Generate and bind corresponding follow-up review periods for different risk grading characteristics, including: generate and bind corresponding follow-up review periods for the high-risk plaque characteristics, FFR-CT values ​​and LDL-C values ​​in the risk grading characteristics.

[0010] Preferably, step S5 further includes: According to the pre-bound feature binding relationship, matching and outputting the follow-up review period and review target corresponding to the risk grading feature of the patient; According to the follow-up review cycle, the patient's CTA images and clinical examination data are collected regularly, and the plaque reversal strategy AI recommendation model is used to identify and determine whether the patient's risk grading characteristics are reduced: If yes, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; Otherwise, continue with the above steps.

[0011] On the other hand, a device for recommending an atherosclerotic plaque reversal strategy based on AI evaluation is provided, and the device for recommending an atherosclerotic plaque reversal strategy based on AI evaluation is used to implement the above-mentioned method for recommending an atherosclerotic plaque reversal strategy based on AI evaluation, and the device comprises: A multimodal data input module, used to input the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; A risk grading generation module, used to identify the high-risk plaque features in the patient's CTA image through the plaque reversal strategy AI recommendation model, and output the risk grading features of the corresponding patient in combination with the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data; A personalized intervention plan generation module is used to match and output an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics according to a pre-bound feature binding relationship; and to match and output a follow-up review cycle and review target corresponding to the patient's risk grading characteristics; The follow-up review management module is used to regularly collect the patient's CTA images and clinical examination data according to the follow-up review period, and identify and determine whether the patient's risk grading characteristics are reduced through the plaque reversal strategy AI recommendation model: if so, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; otherwise, continue the above steps.

[0012] On the other hand, an electronic device is provided, comprising: a processor; and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the methods for recommending strategies for reversing atherosclerotic plaques based on AI assessment as described above is implemented.

[0013] 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 methods for recommending strategies for reversing atherosclerotic plaques based on AI assessment. The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: The present invention proposes an individualized atherosclerotic plaque reversal program using an AI algorithm, which realizes comprehensive analysis of multimodal data through an AI large model algorithm, and provides doctors with personalized intervention programs based on dynamic risk grading, thereby reducing acute cardiovascular events and improving patients' long-term prognosis. Based on the AI ​​large model algorithm, dynamic grading of patient risk levels is achieved through analysis of multimodal data, and personalized lipid-lowering drugs and nutritional supplement programs are generated. The system dynamically adjusts the drug combination, giving priority to statins, and Bempedoic Acid is used for statin intolerance. At the same time, the system combines the patient's risk level to formulate a personalized follow-up review cycle, optimizes intervention and follow-up management, and provides hospitals and physical examination institutions with intelligent and personalized atherosclerotic plaque management and treatment programs to achieve the goal of reversing atherosclerotic plaques.

[0014] The present invention uses a large model algorithm and multimodal data to build an intelligent risk grading system, which recommends drugs and nutritional supplements through a large model to achieve reversal of atherosclerotic plaques in patients. It can process patients' CTA images, FFR-CT values, LDL-C levels and clinical data, and perform risk assessment and grading through a trained AI model to generate a personalized intervention plan. Compared with rule-based systems, the AI ​​algorithm of the present invention has higher flexibility and robustness through large-scale data training. It can dynamically predict the risk grading of patients based on their multidimensional characteristics and provide accurate intervention plans to meet the needs of different patients.

[0015] The system is suitable for hospitals and health management institutions, and is used by cardiovascular specialists in risk assessment, personalized treatment plan generation and intelligent follow-up management in the management of atherosclerotic plaques. The system can automatically evaluate the arterial plaque characteristics, FFR-CT values ​​and LDL-C levels of patients with different risk levels, and support doctors' clinical decisions with personalized treatment plans. It is particularly suitable for patients with high-risk plaques or patients who need to dynamically adjust their treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] 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.

[0017] Figure 1 This is a flow chart of a method for recommending an atherosclerotic plaque reversal strategy based on AI assessment provided by an embodiment of the present invention; Figure 2 It is a block diagram of a device for recommending an atherosclerotic plaque reversal strategy based on AI evaluation provided by an embodiment of the present invention; Figure 3 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0019] 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.

[0020] 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.

[0021] In the embodiments of the present invention, sometimes the subscripts such as W 1It may be mistakenly written as a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0022] 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.

[0023] The embodiment of the present invention provides a method for recommending strategies for reversing atherosclerotic plaques based on AI evaluation, which can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The flowchart of the recommended method for reversing atherosclerotic plaque based on AI assessment is shown. The processing flow of the method may include the following steps: S1. Dataset construction: collect and preprocess a large-scale dataset related to atherosclerosis, including CTA images, FFR-CT values, LDL-C values, and clinical history data; S2. Multimodal data processing: using CNN to extract high-risk plaque features in the CTA image, using GNN to analyze the spatial distribution and structural relationship of the high-risk plaque features, and using the self-attention mechanism to capture the corresponding FFR-CT value, LDL-C value and clinical risk factors for the high-risk plaque features; S3. Risk grading generation: generating individualized risk grading features based on the characteristic relationship of the high-risk plaque features described above; S4. Personalized intervention plan generation: generating and binding corresponding individualized plaque reversal strategy treatment plans for different risk grading characteristics, wherein the plans include drug combinations and nutritional supplements corresponding to the risk grading characteristics; S5. Model training and application: Input the risk grading features into a preset machine learning model for model training to generate a corresponding AI recommendation model for plaque reversal strategy. After clinical verification, the AI ​​recommendation model for plaque reversal strategy is deployed in the background and put into use.

[0024] The present invention mainly realizes the following technical application contents: Risk grading based on AI models: Through large-scale data training of AI models, risk grading based on multimodal patient data is achieved, including risks such as high-risk arterial plaque characteristics (HRP), FFR and LDL-C levels, providing high-precision individualized risk assessment.

[0025] Generation of personalized intervention plans: Based on the risk stratification results, the AI ​​algorithm directly generates the optimal combination of drugs and nutritional supplements to suit the patient's specific health status.

[0026] The steps of training and applying the above model are described in detail below, together with the subsequent follow-up steps (dynamic optimization and self-learning: based on the patient’s follow-up data, the AI ​​model can self-learn and optimize so that the treatment plan is dynamically updated according to the patient’s health status).

[0027] This system uses multimodal AI algorithms to fuse and analyze information from different data sources. Through deep learning models and large-scale data training, the system integrates CTA images, FFR-CT values, LDL-C values ​​and clinical history data into a large model to generate individualized risk grading and intervention plans. The main AI algorithms used are: 1. Convolutional neural network (CNN) for image data processing: CNN is suitable for processing CTA image data and can extract detailed features such as the shape, size and location of plaques; 2. Graph neural network (GNN) is used to characterize the spatial distribution and structural relationship of patch features: GNN identifies the spatial relationship of patches in the network topology, which helps to identify the mutual influence between multiple patches; 3. Multi-head self-attention mechanism is used for data fusion and feature weight adjustment: By assigning dynamic weights to different features (such as HRP, FFR value, LDL-C, etc.), the self-attention mechanism can help the model capture key risk factors in multimodal data.

[0028] Based on the above applications, the present invention is based on the AI ​​large model algorithm, and realizes the dynamic classification of patient risk levels through the analysis of multimodal data, and generates personalized lipid-lowering drugs and nutritional supplements. The system dynamically adjusts the drug combination, giving priority to statins, and Bempedoic Acid is used for statin intolerance. At the same time, the system combines the patient's risk level, formulates a personalized follow-up review cycle, optimizes intervention and follow-up management, and provides hospitals and physical examination institutions with intelligent and personalized atherosclerotic plaque management and treatment plans to achieve the goal of reversing atherosclerotic plaques.

[0029] The steps of constructing the AI ​​model of the present invention are described in detail below. The model training can adopt a CNN model, RNN or random forest model, etc., which is not limited in this embodiment.

[0030] 1. Dataset Construction Data sources: Patient data from medical imaging databases, clinical laboratory data, and follow-up centers.

[0031] Data type: covers annotated CTA images, FFR-CT values, LDL-C levels and multiple follow-up records.

[0032] Dataset size: tens of thousands of annotated CTA images and rich follow-up data.

[0033] Data cleaning and annotation: De-noising, enhancement and standardization of image data were performed, and high-risk plaque features were annotated according to the CAD-RADS 2.0 standard.

[0034] 2. Labeling scheme According to the CAD-RADS 2.0 standard, the following annotations are made for high-risk plaque features: Positive remodeling: The vessel dilates at the plaque, and the ratio of the outer vessel diameter to the average proximal and distal diameters is greater than 1.1.

[0035] Low-attenuation plaque: non-calcified plaque with a density less than 30 HU, which is associated with a higher risk of rupture.

[0036] Spotty calcifications: Small calcification spots within the plaque, indicating high activity.

[0037] Napkin ring sign: an area of ​​low attenuation bordering a ring-like structure of higher CT attenuation, suggesting a lipid-core plaque with a thin fibrous cap.

[0038] 3. Detailed description of AI algorithm The core AI algorithm of the present invention realizes comprehensive analysis of multimodal data and individualized risk grading through the following key modules: Multimodal data processing module: Image processing: Convolutional neural network (CNN) is used to extract plaque morphology and features in patients' CTA images, and classification is performed based on the annotated high-risk plaque features (HRP).

[0039] Structural relationship analysis: Graph neural network (GNN) is used to analyze the spatial relationship of high-risk plaque characteristics and the interaction between different plaques. It is more effective in the case of multiple plaques and identifies potential high-risk areas.

[0040] Numerical data processing: For non-imaging data such as FFR and LDL-C values, the algorithm directly assigns dynamic weights through the self-attention mechanism and combines them with high-risk plaque characteristics to form a complete patient risk profile.

[0041] When the corresponding FFR-CT value, LDL-C value and clinical risk factor are captured for the high-risk plaque feature through the self-attention mechanism, it is necessary to capture the corresponding clinical risk factor for the labeled object of the high-risk plaque feature through the self-attention mechanism.

[0042] Clinical risk factors come from clinical history data, so it is necessary to parse the clinical history data and use the self-attention mechanism to capture clinical risk factors related to the labeled objects of high-risk plaque characteristics "positive reconstruction, low attenuation plaque, napkin ring sign, and macular calcification" (including clinical risk factors, FFR-CT values, and LDL-C values ​​of corresponding characteristics). In this way, the corresponding FFR-CT values, LDL-C values, and clinical risk factors are captured for the high-risk plaque characteristics through the self-attention mechanism; based on the characteristic relationship of the above-mentioned high-risk plaque characteristics, individualized risk grading features (different object characteristics and their clinical risk factors, FFR-CT values, and LDL-C values) are generated.

[0043] 4. Risk classification generation: Data fusion and self-attention mechanism: A multi-head self-attention mechanism is used to assign weights to different features of multimodal data, and to generate individualized risk levels by capturing the relationships between HRP, FFR, and LDL-C features.

[0044] ‌ The following steps can be used to generate individualized risk levels using the multi-head self-attention mechanism:‌ Data Preparation ‌Multimodal Data‌: Collect data containing features such as HRP (heart rate variability), FFR (fractional flow reserve), and LDL-C (low-density lipoprotein cholesterol).

[0045] ‌Feature representation‌: Each feature is represented as a vector to form the input sequence.

[0046] Application of multi-head self-attention mechanism‌ ‌Define input‌: Combine the multimodal data feature vectors into an input sequence and input it into the multi-head self-attention mechanism‌.

[0047] Calculate weights: ‌Query, Key, Value‌: Convert the input sequence into query, key, and value vectors through a learnable parameter matrix.

[0048] ‌Attention head‌: The self-attention calculation process is decomposed into multiple heads, each head independently calculates the attention weight to capture the relationship between different features‌.

[0049] ‌Generate attention matrix‌: In each attention head, calculate the dot product between the query and the key, and apply the softmax function to generate attention weights to represent the correlation between features‌.

[0050] ‌Weighted summation‌: Use the attention weights to perform weighted summation on the value vector to obtain the output of each attention head, which contains the relationship information between features. ‌ ‌Concatenated output‌: Concatenate the outputs of multiple attention heads to obtain a representation containing rich feature interaction information‌.

[0051] ‌Risk score‌: The concatenated feature representations are processed through a fully connected layer or other machine learning models to generate an individualized risk level‌.

[0052] Optimization and Adjustment: According to the actual application effect, the model parameters are adjusted and optimized to improve the accuracy and reliability of risk level generation.

[0053] Through the above steps, the multi-head self-attention mechanism can effectively assign weights to different features of multimodal data and capture the relationships between features to generate individualized risk levels.

[0054] Risk prediction: Based on the data fusion results, the AI ​​model outputs extremely high risk, high risk, moderate risk, and low risk levels, taking into account the patient's plaque characteristics, FFR value, and LDL-C level.

[0055] 5. Personalized intervention plan generation module: Recipe Generation: Use reinforcement learning algorithms to dynamically generate personalized treatment plans based on patient risk stratification, including drug combinations and nutritional supplements.

[0056] Generating and binding corresponding individualized plaque reversal strategy treatment plans for different risk grading characteristics, wherein the plans include drug combinations and nutritional supplements corresponding to the risk grading characteristics; Specifically: The AI ​​model comprehensively considers the patient's plaque characteristics, FFR value and LDL-C level to output extremely high risk, high risk, medium risk and low risk levels. The system configures corresponding personalized intervention plans for different risk characteristics (that is, individualized plaque reversal strategy treatment plans).

[0057] The system uses AI models to recommend appropriate drug combinations and nutritional supplements based on the patient's risk classification, LDL-C level, FFR value, HRP characteristics and other health indicators. The model is trained through deep learning and reinforcement learning technology and has the following characteristics: Very high risk combination: Based on the patient's very high risk classification, the system prioritizes the combination of high-dose statins, ezetimibe, and PCSK9 inhibitors, with the highly effective prescription-grade omega-3 fatty acid EPA. For patients who are intolerant to statins, the model will automatically select Bempedoic Acid as an alternative drug.

[0058] High-risk combination: Based on the high-risk classification, the system recommends a combination of enhanced doses of statins, ezetimibe and PCSK9 inhibitors, with appropriate amounts of EPA and DHA. If the patient is intolerant to statins, the model automatically switches to the Bempedoic Acid alternative.

[0059] Moderate-risk combination: The model will recommend regular doses of statins and appropriate amounts of EPA; if intolerance is detected in the patient, it will automatically adjust to Bempedoic Acid as a substitute.

[0060] Low-risk combination: The system recommends low-dose statins and standard-dose EPA to maintain a low-risk status and optimize the regimen in a timely manner during the review cycle.

[0061] In this way, drug recommendation schemes no longer rely on fixed rules, but are flexibly adjusted through models, thereby enhancing personalization and dynamic responsiveness.

[0062] The risk grading features constructed based on the above feature engineering contain the feature information associated with the annotated objects of different high-risk plaque features, including the clinical risk factors, FFR-CT values, LDL-C values, and structural relationships of the corresponding objects (as well as the binding relationship of the bound individualized plaque reversal strategy treatment plan). Therefore, it is possible to construct a feature set based on the risk grading features of different objects and use it for model training. The risk grading features are input into the preset machine learning model for model training to generate the corresponding plaque reversal strategy AI recommendation model. After clinical verification, the plaque reversal strategy AI recommendation model is deployed in the background and put into use.

[0063] For example, choose the CNN model as the training model; Each risk classification feature is input into the model for training and learning to generate the plaque reversal strategy AI recommendation model. Then, several clinical data are collected as a validation set to verify the recognition performance of the plaque reversal strategy AI recommendation model, and to verify whether the risk level prediction of the clinical data by the plaque reversal strategy AI recommendation model and the corresponding matching output of the individualized plaque reversal strategy treatment plan are consistent with the prediction of the administrator (attending physician) (there may also be minor adjustments, because it is an auxiliary diagnosis output, and the doctor can finally confirm the final individualized plaque reversal strategy treatment plan).

[0064] The performance verification of the AI ​​recommendation model for plaque reversal strategy usually includes the following methods: accuracy, precision, recall, F1 score, confusion matrix, ROC curve (Receiver Operating Characteristic Curve) and AUC value. Accuracy refers to the proportion of correctly predicted samples to the total samples; precision refers to the proportion of correctly predicted positive samples to predicted positive samples; recall refers to the proportion of correctly predicted positive samples to actual positive samples; F1 score is the harmonic mean of precision and recall; confusion matrix shows a detailed comparison between actual categories and predicted categories; ROC curve is used to show the changes in model performance under different thresholds; AUC value is the area under the ROC curve, which is used to evaluate the model's ability to distinguish.

[0065] The above verification is clinically verified by the administrator, for example, using F1 verification.

[0066] The model file can be loaded and deployed on the background server. The clinical data of the patient's examination can be input into the model later. For example, the patient's file information can be read from the background HIS system, and the patient's clinical examination data can be obtained and imported into the model. The model can identify and predict the corresponding risk level, and further match and output the plaque reversal strategy that matches the patient's risk level and recommend it to the doctor and / or patient to assist diagnosis and improve efficiency.

[0067] Preferably, in step S5, the plaque reversal strategy AI recommendation model is deployed in the background and put into use, including: Inputting the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; Identify the high-risk plaque features in the patient's CTA images through the plaque reversal strategy AI recommendation model, combine the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data, and output the risk grading features of the corresponding patient; According to the pre-bound feature binding relationship, an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics is matched and output.

[0068] The data required for subsequent patient examinations, such as CTA images, FFR-CT values, and LDL-C values, can be checked by the hospital's corresponding examination system and uploaded to the backend by the administrator for storage.

[0069] The backend calls the model to identify the high-risk plaque features of the CTA images in the examination data, as well as the FFR-CT value and LDL-C value of the patient, and predicts and outputs the corresponding risk grading features. Based on the binding relationship with the risk grading features, it outputs an individualized plaque reversal strategy treatment plan that matches the risk grading features. The individualized plaque reversal strategy treatment plan output by the system can be pushed to the doctor's PC or PDA terminal or the patient's APP, etc.

[0070] The present invention is also bound with a corresponding follow-up review plan.

[0071] Preferably, step S4 further includes: Generate and bind corresponding follow-up review periods for different risk grading characteristics, including: generate and bind corresponding follow-up review periods for the high-risk plaque characteristics, FFR-CT values ​​and LDL-C values ​​in the risk grading characteristics.

[0072] Preferably, step S5 further includes: According to the pre-bound feature binding relationship, matching and outputting the follow-up review period and review target corresponding to the risk grading feature of the patient; According to the follow-up review cycle, the patient's CTA images and clinical examination data are collected regularly, and the plaque reversal strategy AI recommendation model is used to identify and determine whether the patient's risk grading characteristics are reduced: If yes, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; Otherwise, continue with the above steps.

[0073] For different risk classification characteristics, the administrator can configure the corresponding follow-up review cycle and review target. 1. Extremely high risk patients: CTA + FFR review cycle: every 6-12 months.

[0074] LDL-C review cycle: every 1-2 months.

[0075] Goal: Ensure that LDL-C is continuously controlled at <1.0 mmol / L, monitor plaque changes at all times, prevent acute events, and reduce the risk of new plaque formation.

[0076] 2. High-risk patients: CTA + FFR review cycle: every 12-18 months.

[0077] LDL-C review cycle: every 1-3 months.

[0078] Goal: Control LDL-C to <1.4 mmol / L, regularly monitor plaque progression, and prevent acute events.

[0079] 3. Moderate-risk patients: CTA + FFR review cycle: every 18-24 months.

[0080] LDL-C review cycle: every 3-6 months.

[0081] Goal: Control LDL-C to <1.8 mmol / L to ensure plaque stability and prevent enlargement.

[0082] 4. Low-risk patients: CTA + FFR review cycle: every 24-36 months.

[0083] LDL-C review cycle: every 6-12 months.

[0084] Goal: Control LDL-C to <2.6 mmol / L and maintain a low-risk state for a long time.

[0085] During the background check, the model can further output the corresponding bound follow-up review cycle and review target based on the predicted and identified risk grading characteristics. The system will generate the corresponding follow-up review cycle based on the above follow-up review cycle and review target and the patient's risk grading to optimize the management and intervention effects.

[0086] The patient is reviewed according to the follow-up review cycle, and the review data is uploaded to the background and the model prediction is performed again. The model identifies and outputs the risk grading characteristics after the review, determines whether the risk grading characteristics have changed, and outputs the corresponding adapted individualized plaque reversal strategy treatment plan and the corresponding changed follow-up review cycle and review goals according to the changed risk grading characteristics.

[0087] Therefore, the model can learn and optimize itself: the algorithm is continuously optimized through the patient's follow-up data, and the drug dosage and follow-up frequency are adjusted according to the patient's efficacy feedback. In this way, the drug recommendation plan no longer relies on fixed rules, but is flexibly adjusted through the model, thereby enhancing personalization and dynamic response capabilities.

[0088] Based on the above patient review data, the model can also perform online learning and self-optimization modules: Online learning: Through real-time feedback, the system automatically updates model parameters to adapt to individual changes.

[0089] Transfer learning: The model is applied to new patients or specific high-risk groups to enhance the adaptability of the model through transfer learning.

[0090] The above scheme has good clinical application effect. After clinical verification, it can intelligently output corresponding strategy recommendations for patients with different risk levels to promote the reversal of atherosclerotic plaques. Several cases are provided below: Example 1: High LDL-C, normal FFR-CT, negative HRP Risk rating: Medium risk Patient condition: LDL-C is higher than the target value, FFR-CT is normal (≥0.80), and HRP is negative.

[0091] Drug combination: Atorvastatin 10-20 mg / day.

[0092] Nutritional supplement: High-potency prescription-grade fish oil EPA and DHA 1000 mg / day.

[0093] Goal and monitoring: LDL-C target 1.8 mmol / L, review CTA + FFR every 18-24 months, and review LDL-C every 3-6 months.

[0094] Example 2: High LDL-C, FFR-CT < 0.80, HRP positive Risk classification: High risk Patient condition: LDL-C was significantly elevated, FFR-CT < 0.80, and HRP was positive.

[0095] Drug combination: rosuvastatin 20 mg / day + PCSK9 inhibitor (evolocumab 140 mg / every two weeks or 420 mg / month; or alirocumab 75 mg / every two weeks or 300 mg / month) + ezetimibe 10 mg / day.

[0096] Nutritional supplement: High-potency prescription-grade fish oil EPA and DHA 2000 mg / day.

[0097] Goals and monitoring: LDL-C target 1.4 mmol / L, review CTA + FFR every 12-18 months, and review LDL-C every 1-3 months.

[0098] Example 3: Statin intolerance, high LDL-C, abnormal FFR-CT, positive HRP Risk classification: Very high risk Patient condition: The patient was intolerant to statins and had muscle damage; LDL-C was significantly elevated, FFR-CT <0.80, and HRP was positive.

[0099] Drug combination: PCSK9 inhibitor (evolocumab 140 mg / every two weeks or 420 mg / month; or alirocumab 75 mg / every two weeks or 300 mg / month) + Bempedoic Acid 180 mg / day + Ezetimibe 10 mg / day.

[0100] Nutritional supplement: High-potency prescription-grade fish oil EPA and DHA 2000 mg / day.

[0101] Goal and monitoring: LDL-C target 1.0 mmol / L, review CTA + FFR every 6-12 months, and review LDL-C every 1-2 months.

[0102] Therefore, the present invention combines the patient's risk level to formulate a personalized follow-up review cycle, optimizes intervention and follow-up management, and provides hospitals and physical examination institutions with intelligent and personalized atherosclerotic plaque management and treatment plans to achieve the goal of reversing atherosclerotic plaques.

[0103] Figure 2 1 is a block diagram of a device for recommending an atherosclerotic plaque reversal strategy based on AI evaluation according to an exemplary embodiment. The device is used for recommending an atherosclerotic plaque reversal strategy based on AI evaluation. Figure 2 The device includes a multimodal data input module 310, a risk classification generation module 320, a personalized intervention plan generation module 330, and a follow-up review management module 340. Among them: A multimodal data input module, used to input the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; A risk grading generation module, used to identify the high-risk plaque features in the patient's CTA image through the plaque reversal strategy AI recommendation model, and output the risk grading features of the corresponding patient in combination with the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data; A personalized intervention plan generation module is used to match and output an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics according to a pre-bound feature binding relationship; and to match and output a follow-up review cycle and review target corresponding to the patient's risk grading characteristics; The follow-up review management module is used to regularly collect the patient's CTA images and clinical examination data according to the follow-up review period, and identify and determine whether the patient's risk grading characteristics are reduced through the plaque reversal strategy AI recommendation model: if so, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; otherwise, continue the above steps.

[0104] Please understand the functions and interactions of each module in combination with the above methods, which will not be elaborated here.

[0105] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the electronic device may include the above Figure 2 The device for recommending strategies for reversing atherosclerotic plaques based on AI evaluation is shown. Optionally, the electronic device 410 may include a first processor 2001 .

[0106] Optionally, the electronic device 410 may further include a memory 2002 and a transceiver 2003 .

[0107] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0108] Combine the following Figure 3 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).

[0109] Optionally, the first processor 2001 can execute 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 .

[0110] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 3 CPU0 and CPU1 are shown in FIG.

[0111] In a specific implementation, as an embodiment, the electronic device 410 may also include multiple processors, such as Figure 3The 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).

[0112] 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.

[0113] 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 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0114] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0115] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 3 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0116] 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 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0117] It should be noted that Figure 3 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.

[0118] In addition, the technical effects of the electronic device 410 can refer to the technical effects of the method for recommending atherosclerotic plaque reversal strategies based on AI evaluation described in the above method embodiment, which will not be repeated here.

[0119] 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.

[0120] 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).

[0121] 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.

[0122] 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.

[0123] 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.

[0124] 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.

[0125] 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.

[0126] 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.

[0127] 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.

[0128] 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.

[0129] 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.

[0130] 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.

[0131] 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 who is familiar with the technical field 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. A method for recommending strategies for reversing atherosclerotic plaques based on AI assessment, characterized in that: The method comprises: S1. Dataset construction: collect and preprocess a large-scale dataset related to atherosclerosis, including CTA images, FFR-CT values, LDL-C values, and clinical history data; S2. Multimodal data processing: using CNN to extract high-risk plaque features in the CTA image, using GNN to analyze the spatial distribution and structural relationship of the high-risk plaque features, and using the self-attention mechanism to capture the corresponding FFR-CT value, LDL-C value and clinical risk factors for the high-risk plaque features; S3. Risk grading generation: generating individualized risk grading features based on the characteristic relationship of the high-risk plaque features described above; S4. Personalized intervention plan generation: generating and binding corresponding individualized plaque reversal strategy treatment plans for different risk grading characteristics, wherein the plans include drug combinations and nutritional supplements corresponding to the risk grading characteristics; S5. Model training and application: Input the risk grading features into a preset machine learning model for model training to generate a corresponding AI recommendation model for plaque reversal strategy. After clinical verification, the AI ​​recommendation model for plaque reversal strategy is deployed in the background and put into use.

2. The method for recommending strategies for reversing atherosclerotic plaques based on AI assessment according to claim 1, characterized in that: In step S2, after extracting the high-risk plaque features in the CTA image using CNN, the following steps are further included: The high-risk plaque features are marked, and the marked objects include: Positive remodeling: the blood vessels dilate at the plaque, and the ratio of the outer vessel diameter to the average diameter of the proximal and distal ends is greater than 1.1; Low-attenuation plaque: non-calcified plaque with a density less than 30 HU, which is associated with the risk of rupture; Spotty calcification: small calcification spots within the plaque, indicating high activity; Napkin ring sign: an area of ​​low attenuation bordering a ring-like structure of higher CT attenuation, suggesting a lipid-core plaque with a thin fibrous cap.

3. The method for recommending strategies for reversing atherosclerotic plaques based on AI assessment according to claim 2, characterized in that: In step S2, the corresponding FFR-CT value, LDL-C value and clinical risk factors are captured for the high-risk plaque features through a self-attention mechanism, including: Parsing the clinical history data, and capturing corresponding clinical risk factors for the labeled objects of the high-risk plaque features through a self-attention mechanism; At the same time, the corresponding FFR-CT value and LDL-C value are captured for the labeled object of the high-risk plaque feature through the self-attention mechanism.

4. The method for recommending strategies for reversing atherosclerotic plaques based on AI assessment according to claim 1, characterized in that: In step S5, the AI ​​recommendation model for plaque reversal strategy is deployed in the background and put into use, including: Inputting the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; Identify the high-risk plaque features in the patient's CTA images through the plaque reversal strategy AI recommendation model, combine the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data, and output the risk grading features of the corresponding patient; According to the pre-bound feature binding relationship, an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics is matched and output.

5. The method for recommending strategies for reversing atherosclerotic plaques based on AI assessment according to claim 4, characterized in that: Step S4 also includes: Generate and bind corresponding follow-up review periods for different risk grading characteristics, including: generate and bind corresponding follow-up review periods for the high-risk plaque characteristics, FFR-CT values ​​and LDL-C values ​​in the risk grading characteristics.

6. The method for recommending strategies for reversing atherosclerotic plaques based on AI assessment according to claim 5, characterized in that: Step S5 also includes: According to the pre-bound feature binding relationship, matching and outputting the follow-up review period and review target corresponding to the risk grading feature of the patient; According to the follow-up review cycle, the patient's CTA images and clinical examination data are collected regularly, and the plaque reversal strategy AI recommendation model is used to identify and determine whether the patient's risk grading characteristics are reduced: If yes, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; Otherwise, continue with the above steps.

7. An atherosclerotic plaque reversal strategy recommendation device based on AI evaluation, the atherosclerotic plaque reversal strategy recommendation device based on AI evaluation is used to implement the atherosclerotic plaque reversal strategy recommendation method based on AI evaluation as claimed in any one of claims 1 to 6, characterized in that: The device comprises: A multimodal data input module, used to input the patient's CTA images and clinical examination data into the plaque reversal strategy AI recommendation model; A risk grading generation module, used to identify the high-risk plaque features in the patient's CTA image through the plaque reversal strategy AI recommendation model, and output the risk grading features of the corresponding patient in combination with the high-risk plaque features and the FFR-CT value and LDL-C value in the clinical examination data; A personalized intervention plan generation module is used to match and output an individualized plaque reversal strategy treatment plan corresponding to the patient's risk grading characteristics according to a pre-bound feature binding relationship; and to match and output a follow-up review cycle and review target corresponding to the patient's risk grading characteristics; The follow-up review management module is used to regularly collect the patient's CTA images and clinical examination data according to the follow-up review period, and identify and determine whether the patient's risk grading characteristics are reduced through the plaque reversal strategy AI recommendation model: if so, output the individualized plaque reversal strategy treatment plan with the reduced risk grading characteristics and lower the follow-up review period and review target according to the reduced risk grading characteristics; otherwise, continue the above steps.

8. 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 6 is implemented.

9. 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 6.

Images