Liver cirrhosis acute variceal bleeding ai treatment decision system

Abstract

The application provides an AI treatment decision system for liver cirrhosis acute variceal hemorrhage, relates to the technical field of treatment decision, extracts a transient characteristic indicator of blood flow pressure from a pressure fluctuation signal, predicts a transient pressure peak of the portal vein blood flow during acute variceal hemorrhage through the transient characteristic indicator and a blood flow dynamic model, predicts a quasi-steady pressure field of the portal vein blood flow during acute variceal hemorrhage according to the gradient distribution of the portal vein blood flow pressure and an initial treatment strategy of acute variceal hemorrhage, performs multi-scale coupling on the transient pressure peak and the quasi-steady pressure field to obtain a coupling characteristic of the portal vein blood flow, and then uses the coupling characteristic of the blood flow pressure to perform hierarchical coupling adjustment on the treatment strategy of liver cirrhosis acute variceal hemorrhage. Based on the above scheme, hierarchical coupling of the treatment strategy of liver cirrhosis acute variceal hemorrhage can be realized, so that the prediction accuracy of the portal vein blood flow pressure during acute variceal hemorrhage can be improved.

Description

Technical Field

[0001] This application relates to the field of treatment decision technology, and more specifically, to an AI treatment decision system for acute varicose bleeding in cirrhosis. Background Technology

[0002] Acute variceal bleeding in cirrhosis is a serious complication of portal hypertension, often caused by rupture of esophageal and gastric varices. Patients may experience hematemesis, melena, dizziness, and shock. Diagnosis relies primarily on medical history, physical examination, and gastroscopy. Treatment includes emergency hemostasis (medications, endoscopic procedures, three-lumen two-balloon tube compression), maintaining blood volume, and infection prevention. Long-term management requires prevention of rebleeding, and TIPS or surgical intervention may be necessary.

[0003] In existing technologies, treatment strategies are typically formulated based on general clinical standards, lacking fine-tuning for individual patient pathological conditions. In the treatment of acute varicose bleeding, traditional methods rely heavily on rough assessments based on medical history and symptoms, failing to fully consider the physiological differences of each patient, such as the specific dynamic changes in portal vein blood flow, pressure fluctuations, and vessel wall stress. This can lead to overly simplistic treatment plans that fail to accurately model blood flow pressure fluctuations and their impact on the vessel wall, thus hindering effective prevention or control of bleeding risks. Therefore, achieving graded coupling of treatment strategies for acute varicose bleeding in liver cirrhosis, thereby improving the accuracy of portal vein blood flow pressure prediction in cases of acute varicose bleeding, has become a challenging problem in the field. Summary of the Invention

[0004] This application provides an AI-based treatment decision system for acute varicose bleeding in cirrhosis, which can realize hierarchical coupling of treatment strategies for acute varicose bleeding in cirrhosis, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose bleeding.

[0005] This application provides an AI-based treatment decision-making system for acute varicose bleeding in liver cirrhosis, the decision-making system comprising:

[0006] The model building module is used to collect three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters in cirrhotic veins, and then construct a hemodynamic model of cirrhotic veins using the three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters.

[0007] The transient feature determination module is used to acquire the portal vein pressure fluctuation signal during the stable period before bleeding, and then extract the transient feature indicator factor of blood flow pressure from the pressure fluctuation signal. The transient feature indicator factor and the three-dimensional reconstruction features in the hemodynamic model are used to predict the transient pressure peak of portal vein blood flow during acute varicose bleeding.

[0008] The steady-state feature determination module is used to extract the gradient distribution of portal vein blood flow pressure in the pre-bleeding steady phase based on the hemodynamic features in the hemodynamic model, and then predict the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding.

[0009] The coupling adjustment module is used to couple the transient pressure peak and the quasi-steady-state pressure field at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding, and then use the coupling characteristics of blood flow pressure to perform graded coupling adjustment of the treatment strategy for acute varicose bleeding in cirrhosis.

[0010] In this embodiment, CT angiography combined with a 3.0T magnetic resonance scanner was used to acquire three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters in the cirrhotic veins.

[0011] In this embodiment, ECG-gated phase-contrast magnetic resonance imaging (ECG-gated) sequences are used to acquire phase-contrast magnetic resonance blood flow parameters in cirrhotic veins.

[0012] In this embodiment, extracting transient characteristic indicator factors of blood flow pressure from the pressure fluctuation signal specifically includes:

[0013] Extract the high-frequency components from the pressure fluctuation signal;

[0014] The transient characteristic indicator factor of blood flow pressure is determined by the high-frequency component.

[0015] In this embodiment, predicting the transient pressure peak of portal vein blood flow during acute varicose bleeding using the transient feature indicator factor and the three-dimensional reconstruction features in the hemodynamic model specifically includes:

[0016] Initialize a neural network-based blood vessel wall stress fatigue model;

[0017] The transient feature indicator is used as a dynamic input feature in the blood vessel wall stress fatigue model;

[0018] The three-dimensional reconstruction features in the hemodynamic model are used as structural parameters in the vascular wall stress fatigue model.

[0019] The pressure peak of portal vein blood flow during acute varicose vein bleeding was predicted using the vascular wall stress fatigue model, and the transient pressure peak of portal vein blood flow during acute varicose vein bleeding was obtained.

[0020] In this embodiment, extracting the gradient distribution of portal vein blood flow pressure during the pre-bleeding stable phase based on hemodynamic features from the hemodynamic model specifically includes:

[0021] To obtain the period of stable portal vein blood flow pressure before bleeding;

[0022] Blood flow pressure at various monitoring points in the portal vein during the steady-state period was obtained from the hemodynamic characteristics of the hemodynamic model;

[0023] The gradient distribution of portal vein blood flow pressure during the pre-bleeding stable period was determined based on all blood flow pressures.

[0024] In this embodiment, predicting the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding specifically includes:

[0025] The fluctuation characteristics of blood flow pressure at various monitoring points in the portal vein during treatment were determined based on the initial treatment strategy for acute varicose bleeding.

[0026] The quasi-steady-state pressure field of portal vein blood flow during acute varicose bleeding is predicted by the fluctuation characteristics and the gradient distribution of blood flow pressure.

[0027] In this embodiment, the coupling characteristics of portal vein blood flow during acute varicose vein bleeding are obtained by multi-scale coupling of the transient pressure peak and the quasi-steady-state pressure field, specifically including:

[0028] To obtain the characteristic contribution of features at different scales to portal vein blood flow pressure during portal vein varicose bleeding;

[0029] By weighting and fusing the transient pressure peak and the quasi-steady-state pressure field using the contribution values ​​of each feature, the coupling characteristics of portal vein blood flow during acute varicose bleeding are obtained.

[0030] In this embodiment, the graded coupling adjustment of the treatment strategy for acute variceal bleeding in cirrhosis using the coupling characteristics of blood flow pressure specifically includes:

[0031] Obtain a gradation mapping table for treatment strategies of acute varicose bleeding in liver cirrhosis;

[0032] Treatment levels for acute varicose bleeding in cirrhosis are screened from the level mapping table based on the coupling characteristics of the blood flow pressure.

[0033] The treatment strategy corresponding to the treatment level is used as the treatment strategy for acute varicose bleeding in cirrhosis.

[0034] In this embodiment, the hemodynamic model is a three-dimensional vascular geometric model based on fluid dynamics.

[0035] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0036] By acquiring three-dimensional reconstruction data and phase-contrast magnetic resonance (PCM) blood flow parameters from cirrhotic veins, a hemodynamic model of the cirrhotic veins is constructed. Pressure fluctuation signals of the portal vein during the stable pre-bleeding period are obtained, and transient characteristic indicators of blood flow pressure are extracted from these signals. The transient characteristic indicators and the three-dimensional reconstruction features in the hemodynamic model are used to predict the transient pressure peak of portal vein blood flow during acute varicose bleeding. Based on the hemodynamic features in the hemodynamic model, the gradient distribution of portal vein blood flow pressure during the stable pre-bleeding period is extracted. Then, based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose bleeding, the quasi-steady-state pressure field of portal vein blood flow during acute varicose bleeding is predicted. The transient pressure peak and the quasi-steady-state pressure field are coupled at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding. These coupling characteristics of blood flow pressure are then used to hierarchically couple and regulate the treatment strategy for acute varicose bleeding in cirrhosis.

[0037] Therefore, this application firstly couples transient pressure peaks and quasi-steady-state pressure fields at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding. Then, it uses these blood flow pressure coupling characteristics to perform graded coupling adjustment of treatment strategies for acute varicose bleeding in cirrhosis. Firstly, determining the transient pressure peak helps identify the critical moment of blood flow pressure abrupt changes, reflecting the stress state of the blood vessel during rapid pressure changes. Accurate prediction of transient pressure peaks provides a precise parameter basis for graded coupling of treatment strategies, enabling clinicians to adjust treatment intensity based on changes in blood flow pressure, avoiding over-intervention or under-treatment, and reducing the patient's bleeding risk. By combining individualized transient pressure data, the accuracy of graded adjustment of treatment strategies is improved, providing patients with more targeted treatment plans, thereby effectively improving portal vein blood flow during acute varicose bleeding. The accuracy of portal vein blood flow pressure prediction is improved. Furthermore, determining the quasi-steady-state pressure field helps understand the long-term stability and trends of blood flow during acute varicose vein bleeding, providing a basis for treatment decisions. Analysis of the quasi-steady-state characteristics of the pressure field can reveal the stability and potential risks of blood flow, providing guidance for further treatment strategy adjustments. Accurate prediction of the quasi-steady-state pressure field allows for timely adjustments to treatment intensity when the patient's blood flow pressure reaches a stable or quasi-steady state, avoiding excessive intervention or adverse reactions. This facilitates subsequent multi-scale coupling of the quasi-steady-state pressure field with transient pressure peaks, yielding more precise blood flow coupling characteristics. This provides an effective basis for the graded adjustment of treatment strategies for acute varicose vein bleeding, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose vein bleeding and providing strong support for personalized and optimized treatment plans.

[0038] In summary, the above-mentioned scheme can realize the hierarchical coupling of treatment strategies for acute varicose bleeding in liver cirrhosis, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose bleeding. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a module structure diagram of the AI ​​treatment decision system for acute varicose bleeding in liver cirrhosis provided in this application;

[0041] Figure 2 This is an exemplary flowchart for determining transient pressure peak values ​​according to the present application;

[0042] Figure 3 This is an exemplary flowchart for determining gradient distribution according to the present application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] This application provides an AI-based treatment decision system for acute varicose bleeding in liver cirrhosis. Its core is to extract transient characteristic indicators of blood flow pressure from pressure fluctuation signals. Using these transient characteristic indicators and a hemodynamic model, the system predicts the peak transient pressure of portal vein blood flow during acute varicose bleeding. Based on the gradient distribution of portal vein blood flow pressure and the initial treatment strategy for acute varicose bleeding, the system predicts the quasi-steady-state pressure field of portal vein blood flow during acute varicose bleeding. The system then couples the peak transient pressure and the quasi-steady-state pressure field at multiple scales to obtain the coupling characteristics of portal vein blood flow. These coupling characteristics are then used to hierarchically couple and adjust the treatment strategy for acute varicose bleeding in liver cirrhosis. Based on this scheme, hierarchical coupling of the treatment strategy for acute varicose bleeding in liver cirrhosis can be achieved, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose bleeding.

[0045] To better understand the above technical solutions, a detailed description of the technical solutions will be provided below in conjunction with the accompanying drawings and specific embodiments. (Refer to...) Figure 1 As shown in the figure, this is a module structure diagram of an AI treatment decision system for acute varicose bleeding in liver cirrhosis according to this embodiment of the present application. The decision system includes: a model building module 100, a transient feature determination module 200, a steady-state feature determination module 300, and a coupling adjustment module 400, which are described below:

[0046] The model building module 100 is used to collect three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters in cirrhotic veins, and then construct a hemodynamic model of cirrhotic veins using the three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters.

[0047] It should be noted that in this application, the three-dimensional reconstruction data represents the three-dimensional geometric information of the vascular structure in the cirrhotic veins; the phase-contrast magnetic resonance blood flow parameters represent hemodynamic parameters such as blood flow velocity, flow rate, and shear stress obtained by phase-contrast magnetic resonance imaging (PC-MRI); in specific implementation, CT angiography combined with a 3.0T magnetic resonance scanner is used to acquire the three-dimensional reconstruction data in the cirrhotic veins, and ECG-gated phase-contrast magnetic resonance imaging sequence is used to acquire the phase-contrast magnetic resonance blood flow parameters in the cirrhotic veins.

[0048] In practice, the hemodynamic model of cirrhotic veins can be constructed using 3D reconstruction data and phase-contrast magnetic resonance blood flow parameters in the following manner: A high-precision 3D vascular geometry model is reconstructed using 3D reconstruction data, and phase-contrast magnetic resonance blood flow parameters are used as the fluid boundary conditions of the model. During the model solution process, the Navier-Stokes equations are used to describe blood flow characteristics, combined with a non-Newtonian fluid model (e.g., the Carreau-Yasuda model) to more accurately simulate the hemodynamic characteristics of cirrhotic patients. Numerical calculations are performed using the finite element method or finite volume method, and key parameters such as blood flow velocity, shear stress, and pressure distribution are solved using CFD simulation software (e.g., COMSOL Multiphysics). To improve computational accuracy, mesh optimization is required, and an adaptive mesh refinement strategy is adopted to improve the computational accuracy in vascular stenosis or high shear stress regions. Simultaneously, experimental data (e.g., intravascular pressure data) are used to verify and correct the model, adjusting hemodynamic parameters to optimize simulation accuracy and ensure that the model accurately reflects the true hemodynamic characteristics of cirrhotic veins.

[0049] The transient feature determination module 200 is used to acquire the portal vein pressure fluctuation signal during the stable period before bleeding, and then extract the transient feature indicator factor of blood flow pressure from the pressure fluctuation signal. The transient feature indicator factor and the three-dimensional reconstruction features in the hemodynamic model are used to predict the transient pressure peak of portal vein blood flow during acute varicose bleeding.

[0050] It should be noted that in this application, the pressure fluctuation signal reflects the pulsating nature of blood flow in the vascular system. Specifically, obtaining the portal vein pressure fluctuation signal during the stable pre-bleeding period can be achieved as follows: the Womersley equation can be used to convert preoperative catheter pressure measurement data from transjugular intrahepatic portosystemic shunt (TIPS) into a portal vein pressure fluctuation signal during the stable pre-bleeding period. The Womersley equation is a mathematical model describing periodic fluid flow, particularly suitable for pulsating blood flow in blood vessels. Based on the Navier-Stokes equation, this equation considers the viscosity and pulsating nature of blood and can be applied to a given flow frequency. By considering factors such as vascular geometry and blood viscosity, the Womersley equation can predict the velocity and pressure distribution of blood in blood vessels. Its main function is to introduce frequency domain analysis into fluid dynamics, addressing non-constant blood flow issues. It is particularly suitable for handling pulsating flow and the pressure-velocity relationship in complex vascular systems. Before TIPS, by measuring catheter pressure data, the Womersley equation can be used to convert it into a portal vein pressure fluctuation signal during the pre-bleeding stable phase. This allows for more accurate predictions of blood flow pressure, enabling the assessment of portal vein pressure distribution and its fluctuation characteristics, providing crucial data support for subsequent treatment plans and risk prediction.

[0051] In this embodiment, the extraction of transient characteristic indicator factors of blood flow pressure from the pressure fluctuation signal can be achieved through the following steps:

[0052] Extract the high-frequency components from the pressure fluctuation signal;

[0053] The transient characteristic indicator factor of blood flow pressure is determined by the high-frequency component.

[0054] It should be noted that, in this application, the transient feature indicator is a key parameter reflecting the characteristics of sudden changes in blood flow pressure. In specific implementation, firstly, wavelet transform can be used to extract the high-frequency component in the pressure fluctuation signal, which represents the high-frequency fluctuation component in the blood flow pressure signal. Then, this high-frequency component is used as the transient feature indicator of blood flow pressure. By extracting the high-frequency component in the blood flow pressure signal as the transient feature indicator, the sudden change characteristics of blood flow pressure can be effectively reflected, providing key parameters for predicting transient changes in blood flow and acute events.

[0055] Preferably, in this embodiment, reference Figure 2 As shown, this figure is an exemplary flowchart for determining the transient pressure peak value in an embodiment of this application. In this embodiment, the prediction of the transient pressure peak value of portal vein blood flow during acute varicose vein bleeding using the transient feature indicator factor and the three-dimensional reconstruction features in the hemodynamic model can be achieved by the following steps:

[0056] In step S21, a neural network-based blood vessel wall stress fatigue model is initialized;

[0057] In step S22, the transient feature indicator is used as a dynamic input feature in the blood vessel wall stress fatigue model;

[0058] In step S23, the three-dimensional reconstructed features in the hemodynamic model are used as structural parameters in the vascular wall stress fatigue model;

[0059] In step S24, the pressure peak of portal vein blood flow during acute varicose vein bleeding is predicted using the vascular wall stress fatigue model, and the transient pressure peak of portal vein blood flow during acute varicose vein bleeding is obtained.

[0060] It should be noted that in this application, the transient pressure peak value represents the maximum pressure value reached instantaneously during blood flow pressure fluctuations; the vascular wall stress fatigue model is a mathematical model based on physical mechanics principles, used to describe the long-term stress accumulation and fatigue damage process of the vascular wall under blood flow pressure. This vascular wall stress fatigue model considers the force of blood pulsation on the vascular wall, combining the elasticity of the vascular wall and the stress-strain relationship to simulate the fatigue evolution of the vascular wall under multiple pressure fluctuations. Specifically, the dynamic input features in the vascular wall stress fatigue model, including transient feature indicators reflecting the fluctuation and abrupt changes in blood flow pressure, can influence the stress response of the vascular wall, while the three-dimensional reconstruction features (such as vascular shape and structure) in the hemodynamic model serve as structural parameters, providing geometric information about the vascular wall. Through neural network algorithms, this model can learn complex nonlinear relationships and dynamically predict the stress state of the vascular wall under different pressure fluctuations, especially in acute varicose vein bleeding, providing high-precision prediction of the transient pressure peak value of portal vein blood flow.

[0061] The steady-state feature determination module 300 is used to extract the gradient distribution of portal vein blood flow pressure in the pre-bleeding steady phase based on the hemodynamic features in the hemodynamic model, and then predict the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding.

[0062] Preferably, in this embodiment, reference Figure 3 As shown in the figure, this is an exemplary flowchart for determining the transient pressure peak value in an embodiment of this application. In this embodiment, the gradient distribution of portal vein blood flow pressure in the pre-bleeding stable period can be extracted based on the hemodynamic features in the hemodynamic model using the following steps:

[0063] In step S31, the steady-state period of portal vein blood flow pressure before bleeding is obtained;

[0064] In step S32, the blood flow pressure at each monitoring point in the portal vein during the steady period is obtained from the hemodynamic characteristics of the hemodynamic model;

[0065] In step S33, the gradient distribution of portal vein blood flow pressure during the pre-bleeding stable period is determined based on all blood flow pressures.

[0066] It should be noted that, in this application, gradient distribution represents the rate of change of portal vein blood flow pressure in space; steady period represents the time period during which portal vein blood flow pressure fluctuates less and tends to be constant before bleeding; and blood flow pressure in steady period represents the average pressure value of portal vein blood flow during steady period.

[0067] In practice, firstly, time series analysis methods (e.g., autoregressive moving average (ARMA) models) are used to screen the monitored portal vein blood flow pressure data for stability, thereby identifying time periods with small pressure changes and stable fluctuations, which are defined as the stable period before bleeding. Secondly, the set of all phase-contrast magnetic resonance blood flow parameters in the hemodynamic model is used as the hemodynamic features in the hemodynamic model, and the blood flow parameters of each monitoring point in the stable period of the hemodynamic features of the hemodynamic model are used as the blood flow pressure of each monitoring point in the portal vein during the stable period. Finally, the blood flow pressure difference between adjacent monitoring points can be calculated as the blood flow pressure gradient value, and all blood flow pressure gradient values ​​are arranged according to the location of the monitoring points to represent the gradient distribution of portal vein blood flow pressure during the stable period before bleeding.

[0068] In this embodiment, the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding can be predicted based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding using the following steps:

[0069] The fluctuation characteristics of blood flow pressure at various monitoring points in the portal vein during treatment were determined based on the initial treatment strategy for acute varicose bleeding.

[0070] The quasi-steady-state pressure field of portal vein blood flow during acute varicose bleeding is predicted by the fluctuation characteristics and the gradient distribution of blood flow pressure.

[0071] It should be noted that, in this application, the quasi-steady-state pressure field represents the pressure distribution state of portal vein blood flow that tends to be stable before bleeding but still has small fluctuations; the fluctuation characteristics represent the dynamic characteristics of portal vein blood flow pressure changing over time.

[0072] In practice, firstly, before acute varicose vein bleeding, a hemodynamic simulation model can be used to perform extensive simulations of the initial treatment strategy for acute varicose vein bleeding. The standard deviation of the pressure simulation results corresponding to each monitoring point in the portal vein is used as the fluctuation characteristics of the blood flow pressure at the corresponding monitoring point during treatment, thus obtaining the fluctuation characteristics of the blood flow pressure at each monitoring point in the portal vein during treatment. Then, for each monitoring point in the portal vein, the product of the fluctuation characteristics of the monitoring point and the blood flow pressure gradient value between the monitoring point and the previous monitoring point in the gradient distribution of blood flow pressure is used as the quasi-steady-state pressure prediction value of the monitoring point. In this way, the quasi-steady-state pressure prediction values ​​of each monitoring point in the portal vein can be obtained. Then, all the quasi-steady-state pressure prediction values ​​are arranged in a field according to the location of the monitoring points and used as the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding.

[0073] The coupling adjustment module 400 is used to couple the transient pressure peak and the quasi-steady-state pressure field at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding, and then use the coupling characteristics of blood flow pressure to perform graded coupling adjustment of the treatment strategy for acute varicose bleeding in cirrhosis.

[0074] In this embodiment, the coupling characteristics of portal vein blood flow during acute varicose vein bleeding can be obtained by multi-scale coupling of the transient pressure peak and the quasi-steady-state pressure field using the following steps:

[0075] To obtain the characteristic contribution of features at different scales to portal vein blood flow pressure during portal vein varicose bleeding;

[0076] By weighting and fusing the transient pressure peak and the quasi-steady-state pressure field using the contribution values ​​of each feature, the coupling characteristics of portal vein blood flow during acute varicose bleeding are obtained.

[0077] It should be noted that in this application, coupling features represent the combined influence of transient pressure peak and quasi-steady-state pressure field at multiple scales; feature contribution represents the relative influence weight of features at different scales on portal vein blood flow pressure changes.

[0078] In practice, firstly, historical experience combined with the physical characteristics of cirrhotic patients can be used to predetermine the characteristic contribution of transient and steady-state characteristic scales to portal vein blood flow pressure during varicose vein bleeding. Then, the characteristic contribution of the transient characteristic scale to portal vein blood flow pressure during varicose vein bleeding is used as the fusion weight of the transient pressure peak, and the characteristic contribution of the steady-state characteristic scale to portal vein blood flow pressure during varicose vein bleeding is used as the fusion weight of the quasi-steady-state pressure field. Thus, a Bayesian optimized weight fusion method is used to fuse the transient pressure peak and the quasi-steady-state pressure field. The fused characteristic value of the fused blood flow pressure can be used as the coupling characteristic of portal vein blood flow during acute varicose vein bleeding.

[0079] In this embodiment, the graded coupling adjustment of the treatment strategy for acute variceal bleeding in cirrhosis using the coupling characteristics of blood flow pressure can be achieved through the following steps:

[0080] Obtain a gradation mapping table for treatment strategies of acute variceal bleeding in liver cirrhosis;

[0081] Treatment levels for acute varicose bleeding in cirrhosis are screened from the level mapping table based on the coupling characteristics of blood flow pressure.

[0082] The treatment strategy corresponding to the treatment level is used as the treatment strategy for acute varicose bleeding in cirrhosis.

[0083] In specific implementation, firstly, a level mapping table of treatment strategies for acute varicose bleeding in cirrhosis can be obtained from the console of the treatment decision system; then, the pressure range interval of the coupling characteristics of blood flow pressure in the level mapping table is obtained, and the treatment level corresponding to the pressure range interval is used as the treatment level for acute varicose bleeding in cirrhosis; finally, the treatment strategy corresponding to the treatment level in the console of the treatment decision system can be used as the treatment strategy for acute varicose bleeding in cirrhosis. Through the above scheme, treatment upgrade can be automatically triggered when the predicted pressure value is greater than the preset treatment strategy threshold.

[0084] Therefore, this application firstly couples transient pressure peaks and quasi-steady-state pressure fields at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding. Then, it uses these blood flow pressure coupling characteristics to perform graded coupling adjustment of treatment strategies for acute varicose bleeding in cirrhosis. Firstly, determining the transient pressure peak helps identify the critical moment of blood flow pressure abrupt changes, reflecting the stress state of the blood vessel during rapid pressure changes. Accurate prediction of transient pressure peaks provides a precise parameter basis for graded coupling of treatment strategies, enabling clinicians to adjust treatment intensity based on changes in blood flow pressure, avoiding over-intervention or under-treatment, and reducing the patient's bleeding risk. By combining individualized transient pressure data, the accuracy of graded adjustment of treatment strategies is improved, providing patients with more targeted treatment plans, thereby effectively improving portal vein blood flow during acute varicose bleeding. The accuracy of portal vein blood flow pressure prediction is improved. Furthermore, determining the quasi-steady-state pressure field helps understand the long-term stability and trends of blood flow during acute varicose vein bleeding, providing a basis for treatment decisions. Analysis of the quasi-steady-state characteristics of the pressure field can reveal the stability and potential risks of blood flow, providing guidance for further treatment strategy adjustments. Accurate prediction of the quasi-steady-state pressure field allows for timely adjustments to treatment intensity when the patient's blood flow pressure reaches a stable or quasi-steady state, avoiding excessive intervention or adverse reactions. This facilitates subsequent multi-scale coupling of the quasi-steady-state pressure field with transient pressure peaks, yielding more precise blood flow coupling characteristics. This provides an effective basis for the graded adjustment of treatment strategies for acute varicose vein bleeding, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose vein bleeding and providing strong support for personalized and optimized treatment plans.

[0085] In summary, the above-mentioned scheme can realize the hierarchical coupling of treatment strategies for acute varicose bleeding in liver cirrhosis, thereby improving the accuracy of portal vein blood flow pressure prediction during acute varicose bleeding.

[0086] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0088] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. An AI-based treatment decision-making system for acute varicose bleeding in liver cirrhosis, characterized in that, The decision-making system includes: The model building module is used to collect three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters in cirrhotic veins, and then construct a hemodynamic model of cirrhotic veins using the three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters. The transient feature determination module is used to acquire the portal vein pressure fluctuation signal during the stable period before bleeding, and then extract the transient feature indicator factor of blood flow pressure from the pressure fluctuation signal. The transient feature indicator factor and the three-dimensional reconstruction features in the hemodynamic model are used to predict the transient pressure peak of portal vein blood flow during acute varicose bleeding. The steady-state feature determination module is used to extract the gradient distribution of portal vein blood flow pressure in the pre-bleeding steady phase based on the hemodynamic features in the hemodynamic model, and then predict the quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding. The coupling adjustment module is used to couple the transient pressure peak and the quasi-steady-state pressure field at multiple scales to obtain the coupling characteristics of portal vein blood flow during acute varicose bleeding, and then use the coupling characteristics of blood flow pressure to perform graded coupling adjustment of the treatment strategy for acute varicose bleeding in cirrhosis.

2. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, CT angiography combined with a 3.0T magnetic resonance scanner was used to acquire three-dimensional reconstruction data and phase-contrast magnetic resonance blood flow parameters in the veins of cirrhotic liver.

3. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, Phase-contrast magnetic resonance imaging (PCI) sequences were used to acquire phase-contrast magnetic resonance blood flow parameters in cirrhotic veins.

4. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, Extracting transient characteristic indicators of blood flow pressure from the pressure fluctuation signal specifically includes: Extract the high-frequency components from the pressure fluctuation signal; The transient characteristic indicator factor of blood flow pressure is determined by the high-frequency component.

5. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, The prediction of the transient pressure peak of portal vein blood flow during acute varicose bleeding using the transient feature indicator factors and the three-dimensional reconstruction features in the hemodynamic model specifically includes: Initialize a neural network-based blood vessel wall stress fatigue model; The transient feature indicator is used as a dynamic input feature in the blood vessel wall stress fatigue model; The three-dimensional reconstruction features in the hemodynamic model are used as structural parameters in the vascular wall stress fatigue model. The pressure peak of portal vein blood flow during acute varicose vein bleeding was predicted using the vascular wall stress fatigue model, and the transient pressure peak of portal vein blood flow during acute varicose vein bleeding was obtained.

6. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, The gradient distribution of portal vein blood flow pressure during the pre-bleeding stable phase, extracted based on hemodynamic features from the hemodynamic model, specifically includes: To obtain the period of stable portal vein blood flow pressure before bleeding; Blood flow pressure at various monitoring points in the portal vein during the steady-state period was obtained from the hemodynamic characteristics of the hemodynamic model; The gradient distribution of portal vein blood flow pressure during the pre-bleeding stable period was determined based on all blood flow pressures.

7. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, The quasi-steady-state pressure field of portal vein blood flow during acute varicose vein bleeding, based on the gradient distribution of blood flow pressure and the initial treatment strategy for acute varicose vein bleeding, specifically includes: The fluctuation characteristics of blood flow pressure at various monitoring points in the portal vein during treatment were determined based on the initial treatment strategy for acute varicose bleeding. The quasi-steady-state pressure field of portal vein blood flow during acute varicose bleeding is predicted by the fluctuation characteristics and the gradient distribution of blood flow pressure.

8. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, Multi-scale coupling of the transient pressure peak and the quasi-steady-state pressure field yields the coupling characteristics of portal vein blood flow during acute varicose bleeding, specifically including: To obtain the characteristic contribution of features at different scales to portal vein blood flow pressure during portal vein varicose bleeding; By weighting and fusing the transient pressure peak and the quasi-steady-state pressure field using the contribution values ​​of each feature, the coupling characteristics of portal vein blood flow during acute varicose bleeding are obtained.

9. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, The graded coupling adjustment of treatment strategies for acute variceal bleeding in cirrhosis using the aforementioned blood flow pressure coupling characteristics specifically includes: Obtain a gradation mapping table for treatment strategies of acute variceal bleeding in liver cirrhosis; Treatment levels for acute varicose bleeding in cirrhosis are screened from the level mapping table based on the coupling characteristics of blood flow pressure. The treatment strategy corresponding to the treatment level is used as the treatment strategy for acute varicose bleeding in liver cirrhosis.

10. The AI ​​treatment decision-making system for acute varicose bleeding in liver cirrhosis as described in claim 1, characterized in that, The hemodynamic model is a three-dimensional vascular geometric model based on fluid dynamics.

Images