Remote collaborative management system for chronic biliopancreatic diseases

Abstract

The application relates to the technical field of a biliopancreatic chronic disease remote cooperative management system, and particularly discloses a biliopancreatic chronic disease remote cooperative management system. The system comprises a medical image data acquisition unit, an individualized anatomical model construction unit, a biliopancreatic fluid dynamics simulation engine, a virtual intervention simulation unit, a risk early warning analysis unit and a remote cooperative diagnosis and treatment platform; a digital twin model of a biliopancreatic duct system of a patient is constructed, the biliopancreatic fluid flow state is simulated in combination with computational fluid dynamics, and virtual stent implantation and other intervention effect pre-performance are supported, so that identification of occult obstruction and quantitative evaluation of a treatment scheme are realized; the system also supports multi-territory expert cooperative consultation, dynamic model updating and risk grading early warning. Through the technical scheme, a leap from morphological diagnosis to functional diagnosis is realized, and early lesion detection rate, individualized diagnosis and treatment level and remote cooperation efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of medical information technology, specifically relating to a remote collaborative management system for chronic biliary and pancreatic diseases. Background Technology

[0002] With the rapid evolution of digital healthcare and remote collaboration technologies, remote management of chronic diseases of the biliary and pancreatic systems has become an important way to optimize the allocation of medical resources and improve the efficiency of disease prevention and control. Traditional management models mainly rely on the cloud integration of multimodal medical imaging and electronic medical records, aiming to achieve the downward flow and collaboration of expert resources through cross-regional information exchange. However, the biliary and pancreatic systems possess hydrodynamic properties, and their pathological processes often involve complex fluid transport obstacles and pressure gradient fluctuations, posing challenges to the performance of existing systems in areas such as physiological function quantification, dynamic simulation, and accurate diagnosis.

[0003] Achieving a shift from morphological diagnosis to functional diagnosis is a core objective in the field of remote collaborative management. Current remote assessment systems typically focus only on static anatomical features such as gallstone size and the degree of ductal stenosis, lacking the technical means to transform medical images into individualized physiological flow field models. These systems struggle to recreate in real-time the flow patterns and pressure distribution of bile and pancreatic fluid under different physiological loads, preventing medical personnel from remotely detecting hidden obstructions that have not yet caused anatomical deformation but have already resulted in dynamic abnormalities, thus limiting the sensitivity of early warning systems.

[0004] Existing technologies generally face the problem of mismatch between static imaging data and dynamic disease progression, making it difficult to capture the nonlinear hydrodynamic characteristics within the pancreatic and biliary ducts through linear morphological observation. Due to the lack of digital characterization of biofluid interactions, traditional methods cannot quantitatively assess bile excretion kinetic disorders, nor can they accurately simulate efficacy and predict risks before clinical intervention. The lack of virtual simulation capabilities for surgical or interventional treatments during remote collaboration makes it difficult for clinicians to determine the necessity of referrals and the optimal surgical plan in complex cases, increasing uncertainty in the treatment process.

[0005] Therefore, a remote collaborative management system for chronic biliary and pancreatic diseases is needed. Summary of the Invention

[0006] The purpose of this invention is to provide a remote collaborative management system for chronic biliary and pancreatic diseases, which can solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] The remote collaborative management system for chronic biliary and pancreatic diseases includes a medical imaging data acquisition unit, an individualized anatomical model construction unit, a biliary and pancreatic fluid dynamics simulation engine, a virtual intervention simulation unit, a remote collaborative diagnosis and treatment platform, and a risk warning and analysis unit.

[0009] The medical imaging data acquisition unit is configured to acquire the patient's multimodal medical imaging data, including computed tomography images and magnetic resonance imaging images, and extract anatomical information of the bile and pancreatic duct system.

[0010] The individualized anatomical model construction unit is configured to reconstruct a three-dimensional geometric model of the patient's bile and pancreatic duct system based on the anatomical structure information acquired by the medical imaging data acquisition unit, and to assign it physiological boundary conditions to form an individualized digital twin basic model.

[0011] The biliary and pancreatic fluid dynamics simulation engine is configured to treat biliary and pancreatic fluid as a non-Newtonian fluid. On the digital twin basic model generated by the individualized anatomical model building unit, it simulates the flow state, pressure distribution and shear force of biliary and pancreatic fluid under different body position changes and dietary stimuli, and outputs dynamic flow field parameters.

[0012] The virtual intervention simulation unit is configured to implant virtual stents or simulate other interventional procedures based on the flow field established by the pancreatic and biliary fluid dynamics simulation engine, compare the changes in fluid dynamic parameters before and after intervention, and evaluate the potential effects of surgery or interventional treatment.

[0013] The risk warning analysis unit is configured to identify hidden obstruction areas, quantify the degree of dynamic impairment, and predict the risk of stone recurrence or the path of inflammation spread based on the flow field characteristics output by the biliary and pancreatic fluid dynamics simulation engine and the virtual intervention simulation unit, and generate graded warning signals.

[0014] The remote collaborative diagnosis and treatment platform is configured to integrate the data and analysis results of the above-mentioned units, support online collaborative consultations among medical experts from multiple regions, share digital twin models and simulation processes, and formulate individualized treatment strategies based on virtual intervention simulation results.

[0015] Preferably, the individualized anatomical model building unit is further configured to perform fine mesh division of the narrow, dilated, and branch structures of the bile and pancreatic duct system to ensure that the spatial resolution of the fluid simulation meets the needs of clinical diagnosis.

[0016] Furthermore, the bile and pancreatic fluid dynamics simulation engine employs computational fluid dynamics algorithms to set the inlet flow rate, outlet pressure, and tube wall elastic modulus based on the patient's physiological parameters, and dynamically adjusts the fluid viscosity and density during the simulation process to reflect changes in bile composition.

[0017] Furthermore, the virtual intervention simulation unit supports parametric modeling of various virtual instruments, including stent models with different diameters, lengths and materials, and can simulate the wall adhesion effect and disturbance to the surrounding flow field after the stent is released.

[0018] Preferably, the risk warning analysis unit is set with a preset threshold. When the simulated local pressure gradient or shear force exceeds the preset threshold, it is automatically marked as a high-risk area, and the disease progression trend is assessed in combination with historical disease data.

[0019] Furthermore, the remote collaborative diagnosis and treatment platform is equipped with an interactive visual interface, allowing users to rotate, slice, or zoom in on the digital twin model in real time, and simultaneously play fluid simulation animations, enabling remote doctors to intuitively understand the mechanism of abnormal bile and pancreatic fluid flow.

[0020] Furthermore, the medical image data acquisition unit is also configured to receive image updates during patient follow-up and trigger the individualized anatomical model construction unit to dynamically correct the digital twin model, thereby achieving continuous tracking of disease progression.

[0021] Preferably, the bile and pancreatic fluid dynamics simulation engine supports parallel simulation under multiple working conditions, and can simultaneously run flow field calculations under various physiological states such as fasting, postprandial, and changes in body position, and generate comparative analysis reports.

[0022] Furthermore, the virtual intervention simulation unit works in conjunction with the risk warning analysis unit. When the simulation shows that the improvement in flow rate after virtual stent implantation is lower than the predetermined standard, the system automatically suggests prioritizing conservative treatment and generates corresponding evidence for remote consultation reference.

[0023] Furthermore, the remote collaborative diagnosis and treatment platform has a built-in permission management module to ensure that patient data complies with medical information security standards in multi-center collaboration, and supports structured recording and retrospection of diagnosis and treatment opinions.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. This invention utilizes computational fluid dynamics algorithms to perform high-precision simulation of bile and pancreatic fluid flow, enabling remote doctors to "see" hidden dynamic obstruction points, thereby improving the detection capability and diagnostic accuracy of early lesions.

[0026] 2. By visualizing the fluid effects of interventions such as virtual stent implantation, the system provides a quantitative basis for surgical decisions, avoiding unnecessary referrals and invasive procedures, and reducing medical risks and costs.

[0027] 3. The multi-condition simulation and dynamic model update mechanism supports continuous monitoring of disease progression. Combined with risk warning analysis, it forms a closed loop covering the entire cycle of screening, assessment, intervention simulation and follow-up management, which enhances the scientific, forward-looking and individualized level of remote collaborative diagnosis and treatment. Attached Figure Description

[0028] Figure 1This is a schematic diagram of the overall technical solution architecture of the present invention;

[0029] Figure 2 This is a schematic diagram of the core principle framework for the hydrodynamic calculation of bile and pancreatic fluid based on non-Newtonian fluid simulation in this invention.

[0030] Figure 3 This is a logical flowchart of the multimodal medical image data acquisition and personalized digital twin model construction in this invention;

[0031] Figure 4 This is a logical flowchart of the virtual intervention simulation comparative analysis and hidden obstruction risk classification and early warning in this invention;

[0032] Figure 5 This is a schematic diagram of the data flow for multi-center expert interaction and continuous tracking of disease progression under the remote collaborative diagnosis and treatment platform of this invention. Detailed Implementation

[0033] Example 1: Please refer to the appendix Figure 1 To be continued Figure 5 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0034] The remote collaborative management system for chronic biliary and pancreatic diseases includes a medical image data acquisition unit, an individualized anatomical model construction unit, a biliary and pancreatic fluid dynamics simulation engine, a virtual intervention simulation unit, a risk warning analysis unit, and a remote collaborative diagnosis and treatment platform.

[0035] The medical imaging data acquisition unit is used to acquire multimodal medical imaging data of patients and extract anatomical structural information of the biliary and pancreatic duct system. The unit is configured to retrieve raw data, including computed tomography (CT) images and magnetic resonance imaging (MRI) images, from hospital information systems or image archiving and communication systems via standard medical digital imaging and communication protocol interfaces. The unit integrates an image preprocessing subunit configured to perform noise reduction processing on the raw MRI pancreatobiliary imaging data, employing an algorithm based on anisotropic diffusion filtering to eliminate random interference components in the image background. The unit also includes an anatomical feature recognition module, which uses a deep learning convolutional neural network to automatically segment the common bile duct, pancreatic duct, gallbladder, and duodenal papilla region, determine the spatial topological connections of the biliary and pancreatic duct system, and extract anatomical structural parameters including duct diameter, bending radius, and branching angle.

[0036] The individualized anatomical model construction unit, connected to the medical imaging data acquisition unit, is used to reconstruct a three-dimensional geometric model of the patient's bile and pancreatic duct system and form an individualized digital twin base model. The individualized anatomical model construction unit includes a three-dimensional surface reconstruction submodule, which is configured to use a moving cube algorithm to convert the extracted anatomical structure information into geometric surfaces represented by triangular facets. The individualized anatomical model construction unit also includes a geometry cleaning and optimization submodule, used to eliminate small holes or singular facets generated during reconstruction, ensuring the topological manifold characteristics of the model. The individualized anatomical model construction unit is further configured to perform refined mesh generation on the narrow, dilated, and branching structures of the bile and pancreatic duct system. Specifically, the unit has a built-in mesh generator capable of generating adaptive unstructured tetrahedral meshes for complex anatomical locations such as the distal common bile duct, and densifying the boundary layer of the duct wall by setting at least five proportionally scaled prism elements to accurately capture the velocity gradient near the wall. The individualized anatomical model construction unit is equipped with a physical property definition interface for assigning physiological boundary conditions to the model. The physiological boundary conditions include a function of the inlet flow rate changing with time, the outlet pressure hydrostatic pressure, and the linear or nonlinear elastic modulus of the tube wall, which are set according to the patient's individualized examination results.

[0037] The biliary and pancreatic fluid dynamics simulation engine, connected to the individualized anatomical model construction unit, is used to simulate the flow state of biliary and pancreatic fluid on a digital twin-based model. The engine employs computational fluid dynamics algorithms, treating biliary and pancreatic fluid as a non-Newtonian fluid with shear-thinning properties and selecting an appropriate constitutive model for characterization. The engine includes a governing equation solving module configured to discretize and solve the continuity equation and the Navier-Stokes equations in three-dimensional space. The density of biliary and pancreatic fluid is set as a variable that dynamically changes with bile concentration, and the dynamic viscosity is described as a function of shear rate. The velocity vector and pressure scalar of each grid cell at a specific moment are obtained through iterative calculation.

[0038] The biliary and pancreatic fluid dynamics simulation engine is further configured to support parallel simulation under multiple operating conditions, including a body position change simulation subunit and a dietary stimulation simulation subunit. The body position change simulation subunit is configured to dynamically adjust the component direction of the gravity vector in the three-dimensional coordinate system according to whether the patient is in a supine, lateral, or standing position, simulating the deposition patterns of bile in the gallbladder and the obstruction of its excretion into the common bile duct under different gravity environments. The dietary stimulation simulation subunit is configured to adjust the flow pulse function at the inlet of the common bile duct according to a preset postprandial physiological response model, simulating transient flow field changes during gallbladder contraction and sphincter relaxation. The dynamic flow field parameters output by the biliary and pancreatic fluid dynamics simulation engine include a three-dimensional streamline diagram, a pressure gradient distribution diagram, the time-averaged shear force of the duct wall, and local eddy current intensity. These parameters are transmitted to subsequent units in the form of a structured data stream.

[0039] The virtual intervention simulation unit is used to simulate interventional procedures and evaluate their effects based on the flow field established by the simulation engine. The virtual intervention simulation unit includes a virtual instrument library module, which stores various parametrically modeled stent models. Each stent model is associated with physical property parameters, including stent diameter, length, radial support force, and porosity after expansion. The virtual intervention simulation unit has an intervention path planning submodule, allowing physicians to select the stent implantation location and coverage area in virtual space. The virtual intervention simulation unit is configured to perform geometric deformation calculations to simulate the topological changes to the original three-dimensional geometric model of the biliary and pancreatic duct system after stent implantation, such as simulating the mechanical expansion effect of the stent on stenotic sites. After the geometric morphology is updated, the virtual intervention simulation unit triggers the biliary and pancreatic fluid dynamics simulation engine to perform secondary calculations to obtain the post-intervention flow field distribution. The virtual intervention simulation unit also includes a comparative evaluation subunit, which is configured to calculate the pressure loss improvement rate before and after intervention, i.e., the percentage obtained by subtracting the proximal and distal pressure difference before intervention from the proximal and distal pressure difference after intervention, and then dividing by the pressure difference before intervention; at the same time, it calculates the flow velocity distribution in the stent and the edge region to assess whether there is a risk of thrombosis or deposit formation due to fluid retention.

[0040] The risk warning analysis unit is used to identify obstruction areas and generate warning signals. It incorporates risk quantification and evaluation logic, receiving flow field characteristic data from the biliary and pancreatic fluid dynamics simulation engine and the virtual intervention simulation unit. The unit includes a hidden obstruction identification module, which uses a pressure gradient anomaly detection algorithm to automatically identify hidden obstruction points when the instantaneous pressure gradient value of a local area exceeds a preset physiological threshold for multiple consecutive cardiac or peristaltic cycles. The unit further includes a dynamic obstacle quantification module, configured to calculate the bile emptying fraction, which is the ratio of the total fluid volume passing through the distal end of the common bile duct to the theoretical discharge volume at the inlet end during a complete physiological cycle.

[0041] The risk warning analysis unit is equipped with a tiered warning logic: when the calculated local maximum shear force exceeds a first preset warning value and the bile emptying fraction is lower than a second preset warning value, the system generates a level-one high-risk signal, indicating a risk of acute stone obstruction or acute inflammation; when only the local pressure gradient is high, but overall emptying is normal, a level-two medium-risk signal is generated, indicating the possibility of chronic inflammation progression or potential microstone formation. The risk warning analysis unit also includes a disease course prediction submodule, which is configured to use historical follow-up data to construct a time-series analysis model to predict the probability of further narrowing of the stenotic segment diameter and the geometric growth trend of stone recurrence within a specific future time period without intervention, generating a prediction report including a risk path map.

[0042] The remote collaborative diagnosis and treatment platform is used to integrate data and support online consultations. The platform is equipped with a data integration server configured to uniformly manage patients' original images, digital twin models, simulation animations, and risk reports. The platform features an interactive visualization interface built on a 3D graphics rendering engine, supporting cross-platform remote access. This interactive visualization interface is configured to perform the following functions: support remote experts in rotating, sectioning, or zooming in on the 3D digital twin model in real time via terminal devices; support synchronous playback of fluid dynamic simulation videos of bile and pancreatic fluid flow, displaying real-time changes in pressure or shear force in the form of color contour lines.

[0043] The remote collaborative diagnosis and treatment platform also includes a collaborative consultation management submodule, which is configured to support simultaneous online access by multiple medical experts from various centers and provides voice, text, and whiteboard annotation functions. Experts can directly perform measurements and annotations on the 3D model, and the relevant annotation information is synchronized to all participating terminals in real time. The remote collaborative diagnosis and treatment platform has a built-in strategy formulation assistant, which is configured to automatically rank the optimal intervention strategy recommendations based on the comparison results of multiple solutions output by the virtual intervention simulation unit, for the expert group's decision-making reference. The platform has an access control and security audit module, adopting a role-based access control model to ensure that the transmission and storage of medical data comply with medical information security standards, and to store all operation records in a structured manner to achieve the traceability of diagnosis and treatment opinions.

[0044] The medical imaging data acquisition unit also features continuous monitoring logic, configured to periodically receive new imaging data during patient follow-up. Upon receiving updated images, the unit automatically triggers the individualized anatomical model construction unit to incrementally correct the existing digital twin model. Through image registration technology, it compares the geometric differences between the two periods, calculating the change in stenosis rate or the impact of newly formed space-occupying lesions. The biliary and pancreatic fluid dynamics simulation engine then re-executes the simulation task based on the corrected model, achieving continuous dynamic tracking of the evolution of chronic biliary and pancreatic diseases.

[0045] In this embodiment, the bile and pancreatic fluid dynamics simulation engine specifically considers the influence of bile components on the flow field when performing non-Newtonian fluid simulations. The system allows users to input bile biochemical indicators, such as total bilirubin concentration and cholesterol saturation. The engine internally stores a viscosity correction function, which is configured to nonlinearly correct the characteristic viscosity coefficient of the fluid based on the deviation of cholesterol saturation from a preset benchmark value, so that the simulation process can more realistically reflect the contribution of bile stasis or changes in bile properties to flow resistance.

[0046] The virtual intervention simulation unit further considers the mechanical interaction between the stent and the bile duct wall during stent implantation simulation. The unit includes a contact mechanics analysis submodule, configured to calculate the circumferential stress distribution on the bile duct wall after stent release. If the calculated local stress exceeds the elastic limit of the bile duct tissue, the system automatically issues a structural damage warning and suggests reducing the nominal diameter of the virtual stent or replacing it with a stent model made of a softer material, thus avoiding iatrogenic injury during the simulation phase.

[0047] The risk warning analysis unit is also configured to combine the patient's clinical biochemical indicators when generating warning signals. The unit includes a feature fusion assessment module, which weights and fuses kinetic parameters obtained from computational fluid dynamics with laboratory indicators such as serum amylase and alkaline phosphatase, and uses a logistic regression model or Bayesian network to derive a comprehensive risk score. When the comprehensive risk score reaches a predetermined high threshold, the remote collaborative diagnosis and treatment platform automatically pushes a notification to the attending physician via mobile device and automatically initiates a multidisciplinary expert consultation process.

[0048] The interactive visualization interface of the remote collaborative diagnosis and treatment platform is also equipped with a virtual endoscopy browsing function. This function is configured to generate simulated endoscopic images using a camera roaming algorithm, following the flow path within the bile and pancreatic ducts. Remote physicians can then view the three-dimensional ductal system as if operating a real endoscope, directly observing the luminal morphology of narrowed areas and the complex flow structure of bile at branch confluences. This visualization method effectively compensates for the shortcomings of traditional two-dimensional cross-sectional images in representing spatial continuity.

[0049] Example 2: This example describes a remote collaborative management system for chronic biliary and pancreatic diseases based on a distributed edge computing architecture. Unlike the centralized architecture of Example 1, this example emphasizes performing high-load fluid dynamics calculations locally at the medical institution to reduce latency in transmitting ultra-large-scale 3D model data over the public network, while also improving the system's disaster recovery capabilities.

[0050] The remote collaborative management system for chronic diseases of the biliary and pancreatic system includes edge computing nodes deployed within the hospital's local area network and a central management platform deployed in the cloud.

[0051] The edge computing node includes a local medical image access unit, a local anatomical model generation unit, a distributed dynamics simulation module, and a data synchronization proxy unit.

[0052] The local medical imaging access unit is used to extract raw slice images of specific patients from the hospital's internal image database. Because this unit is located within a local area network, it can quickly acquire high-resolution thin-slice scan data using high-speed bandwidth. The unit has a built-in image quality assessment model configured to automatically detect artifacts caused by the patient's respiratory movements during the data acquisition phase. If the artifact level exceeds a preset signal-to-noise ratio threshold, it automatically issues a retransmission request or prompts the radiologist to reconstruct the image sequence.

[0053] The local anatomical model generation unit is used to perform rapid 3D reconstruction based on locally accessed image data using the graphics processing unit (GPU) cluster of the edge server. The unit employs a hybrid mesh generation algorithm, configured to divide the main trunk of the biliary and pancreatic duct system into a structured hexahedral mesh, while filling irregular areas such as the ends of branches with an unstructured tetrahedral mesh. This hybrid meshing method, while maintaining computational accuracy, reduces the total number of elements compared to a purely unstructured mesh, thus improving the speed of subsequent simulations.

[0054] The distributed dynamics simulation module is configured to utilize the multi-core parallel computing capabilities of edge servers to perform numerical solutions for the bile and pancreatic fluid flow field. The module integrates a specific hardware acceleration library for computational fluid dynamics, using block-based cyclic distribution storage of large sparse matrices in the governing equations and parallel communication interfaces to achieve data exchange across processor cores. During the simulation, the module is configured to monitor the residual convergence in real time. If the residual decrease rate falls below a preset extreme value for 10 consecutive iterations, the time step is automatically adjusted or a higher-order discretization scheme is adopted to ensure the numerical stability of the flow field simulation under complex and narrow structures.

[0055] The data synchronization proxy unit is responsible for desensitizing and extracting features from the locally generated preliminary simulation results. This unit is configured to transform the massive, complete grid data into key feature point cloud data and to lightweight encapsulate key physical quantities in the flow field (such as peak pressure and average shear force). The data synchronization proxy unit uses an encrypted secure tunnel to transmit these lightweight, individualized digital twin feature data to the central management platform in the cloud in real time.

[0056] The cloud-based central management platform includes a virtual intervention global simulator, a risk collaborative assessment unit, and an expert interaction portal.

[0057] The virtual intervention global simulator receives feature data from multiple edge nodes and supports concurrent virtual surgical drills by multiple users. The simulator is configured to maintain a standardized stent performance database in the cloud. When a remote expert initiates an intervention simulation request, the simulator does not need to rebuild a complete anatomical model. Instead, based on lightweight features synchronized from the edge nodes, it uses alternative model algorithms (such as Kriging interpolation or neural network surrogate models) to quickly estimate the post-intervention flow field response. This surrogate model-based computation method can reduce the feedback time of intervention effects from hours to seconds, meeting the needs of real-time online discussions.

[0058] The risk collaborative assessment unit is used to aggregate similar case data from different medical institutions and construct an early warning logic based on a knowledge graph. This unit is configured to match the current patient's hemodynamic characteristics with tens of thousands of past cases in the database. By calculating Mahalanobis distance or cosine similarity, it identifies the historical cases most similar in geometric shape and flow field distribution, and extracts their final treatment outcomes as a reference for the current early warning. The risk collaborative assessment unit can dynamically adjust the threshold parameters of each level of early warning based on the distribution patterns of similar cases, enabling the system to continuously evolve and learn.

[0059] The expert interaction portal is configured to provide a lightweight 3D view accessible via a browser. Employing graphics streaming technology, high-quality 3D fluid animations rendered in the cloud are pushed to the experts' terminal devices via video stream, ensuring a smooth interactive experience even in mobile environments with limited network connectivity. The portal includes a built-in collaborative note-taking system that can structurally link revisions, manual annotations, and virtual intervention parameters from experts at different centers to create a complete collaborative diagnosis and treatment report.

[0060] In this embodiment, a heartbeat-based synchronization status monitoring mechanism is used between the edge computing nodes and the cloud platform. If a hardware failure occurs on the edge computing node during simulation, the data synchronization agent unit immediately switches to the backup computing link and uploads a snapshot of the completed computing status to the cloud. The central management platform in the cloud can take over the remaining computing tasks, ensuring uninterrupted diagnosis and treatment processes through the elastic computing resources of the cloud. The edge computing node is also equipped with a local caching module to store the patient's recent multimodal images and model data. When the same patient undergoes a follow-up examination, the system can perform rapid differential analysis based on the cached data, further improving diagnostic efficiency.

[0061] Example 3: This example describes a remote collaborative management system for chronic biliary and pancreatic diseases with enhanced intelligent analysis capabilities. The system integrates advanced machine learning algorithms to automatically identify complex anatomical variations in the biliary and pancreatic ducts and can automatically optimize the selection of virtual stents based on hydrodynamic characteristics.

[0062] The remote collaborative management system for chronic diseases of the gallbladder and pancreas includes an intelligent sensing and acquisition unit, an adaptive modeling unit, an intelligent fluid simulation engine, an automated intervention design module, a deep risk mining unit, and a multi-terminal collaborative cloud platform.

[0063] The intelligent sensing and acquisition unit, in addition to its conventional image acquisition functions, is also configured as a multi-source heterogeneous data fusion center. This unit includes a natural language processing submodule, capable of automatically capturing and structuring patients' electronic medical record information, including past surgical history, allergy records, and physical examination results. The unit is equipped with data consistency verification logic; when extracted anatomical features (such as common bile duct width) logically conflict with the clinical manifestations described in the medical record, the system automatically marks the point of contention and alerts the doctor on the collaborative diagnosis and treatment platform for manual review.

[0064] The adaptive modeling unit integrates a topology generation model based on anatomical prior knowledge. This unit is configured to use a statistical shape model to assist in inferring areas with poor image quality during the reconstruction of the 3D geometric model. For example, in cases where the distal pancreatic duct is poorly visualized in MRI images, the unit automatically generates a model of the distal structure that best matches the physiological and anatomical probability, based on the trajectory of the pancreatic duct trunk and the spatial constraints of surrounding organs. This adaptive modeling unit possesses mesh quality self-evaluation logic, capable of calculating the Jacobian determinant value of each mesh cell. When the mesh distortion rate exceeds a preset threshold, it automatically triggers a local re-meshing process.

[0065] The intelligent fluid simulation engine is configured as a multi-scale physical model. This adaptive modeling engine also includes a microscopic particle tracking submodule. This submodule is configured to model microscopic stones or precipitates in bile as discrete-phase particles, simulating the trajectories of these particles in the flow field and their deposition probabilities at pipe walls and narrow points. The particles are subjected to a combination of fluid drag, lift, Brownian force, and gravity, and the engine obtains the transient positions of the particles by solving the Langevin equation. This multi-scale simulation can more proactively predict the potential risk of biliary pancreatitis caused by microscopic stones.

[0066] The automated intervention design module replaces the manual selection process. This module is configured to execute optimization logic based on a genetic algorithm: using the minimization of total pressure drop within the bile and pancreatic ducts after intervention and the homogenization of local shear forces as objective functions, it automatically searches for the optimal combination of stent diameter, length, and material stiffness within the stent parameter space. The module can generate a series of Pareto optimal solutions and visualize the performance-risk balance between different options. Furthermore, the module supports sensitivity analysis of stent placement location, automatically identifying key intervention sites most sensitive to flow field improvement.

[0067] The deep risk mining unit is equipped with a prediction model based on a long short-term memory network. This unit is configured to receive time-series flow field data from the simulation engine and analyze the coupling relationship between pressure fluctuation cycles and the patient's physiological rhythms such as respiration and pulse. The deep risk mining unit can identify nonlinear dynamic features that are difficult to detect using traditional thresholding methods, such as abnormal pressure wave reflections or prolonged durations of local reflux zones. The unit combines these deep features with the patient's genomic information or biomarker data to output a higher-dimensional disease progression warning index.

[0068] The multi-terminal collaborative cloud platform is configured to support multi-level linkage between mobile devices, workstations, and large-screen display walls. The platform incorporates intelligent triage logic, which automatically schedules the optimal remote consultation time window based on the urgency of the patient's condition and the expert team's scheduling, using optimization algorithms. The collaborative platform also features a virtual reality (VR) interface, allowing doctors to wear head-mounted displays to enter a virtual bile and pancreatic duct space for immersive examination. In VR mode, doctors can directly adjust the position of the virtual stent through gesture recognition, and the system provides real-time force and tactile feedback, simulating the resistance effect when the stent contacts the duct wall.

[0069] In this embodiment, the intelligent fluid simulation engine also has an automatic tuning function, which can reverse-correct simulation parameters based on real-time feedback of clinically measured pressure data (such as values ​​obtained through bile duct pressure measurement). The engine is configured as a closed-loop feedback system: calculating the residual between the simulated values ​​and the measured values, using the adjoint equation method to solve the sensitivity gradient of the objective function to boundary conditions, and automatically correcting the inflow boundary values ​​or tube wall compliance parameters in the model to ensure a high degree of consistency between the digital twin model and the real physiological state.

[0070] The deep risk mining unit employs a numerical simulation method based on the reaction-diffusion equation when assessing the inflammation spread pathway. This method is configured to treat the concentration distribution of inflammatory mediators as a process influenced by both fluid convection and molecular diffusion, simulating the dynamic pathways of bacteria or inflammatory factors retrogradely entering the pancreatic duct or intrahepatic bile ducts under cholestasis. By calculating the spatial range within which inflammatory mediators exceed pathogenic concentrations over a specific time period, the system can provide clinicians with targeted anti-infective treatment recommendations and provide early warnings of potential sepsis risk points.

[0071] The automated intervention design module further integrates simulations of bile duct dynamics. This module is configured to simulate the effect of spontaneous peristalsis of bile duct smooth muscle on the flow field. By applying traveling wave-like displacement boundary conditions to the surface of the three-dimensional model, the fatigue life of the stent and the probability of stent displacement are calculated under different peristaltic frequencies and amplitudes. If the simulation results indicate a high risk of stent dislodgement, the system will automatically adjust the optimization scheme, suggesting the addition of a stent anchoring structure or the selection of a covering membrane material with a higher coefficient of friction.

[0072] The multi-terminal collaborative cloud platform employs a blockchain-based medical data sharing architecture for data management. Every remote consultation opinion, every generated digital twin version, and every record of a virtual intervention simulation is encapsulated as a timestamped block and distributed across multiple central nodes. This architecture ensures the immutability of the medical decision-making process, providing a solid foundation of data credibility for handling complex medical disputes or multi-center clinical research. The platform supports automated evaluation of treatment quality, automatically calculating the system's diagnostic accuracy and missed diagnosis rate by comparing system alert results with final clinical follow-up results, providing data support for system iteration and upgrades.

[0073] Example 4: This example describes a remote collaborative management system for chronic biliary and pancreatic diseases targeting primary healthcare institutions. The system is designed with lightweight deployment and a simplified operating process in mind, aiming to enable primary hospitals lacking complex computing resources and professional simulation personnel to enjoy high-level digital twin diagnostic and treatment services.

[0074] The remote collaborative management system for chronic biliary and pancreatic diseases includes a primary-level data collection workstation, a cloud-based simulation service cluster, and collaborative terminals.

[0075] The primary care acquisition workstation is configured as an integrated hardware and software unit. This workstation connects directly to the CT / MRI equipment of the primary care hospital via a dedicated data bus. The workstation includes an image compression and intelligent upload module. This module employs a lossy compression algorithm based on regional focus, compressing the original large-capacity medical images to less than 10% of their original size while maintaining the resolution of key anatomical sites in the biliary and pancreatic duct system. This facilitates rapid uploads even in low-bandwidth internet environments. Simultaneously, the workstation features a simplified user interface; medical technicians only need to click to confirm, and the system automatically performs image sequence selection and standardized naming.

[0076] The cloud-based simulation service cluster hosts all the core computing tasks of the system. This cluster is configured as a containerized platform with elastic scaling capabilities. When multiple workstations simultaneously initiate simulation requests, the cluster management system automatically allocates GPU computing resources and launches multiple parallel simulation containers. The cloud cluster contains a large, standardized bile and pancreatic flow field template library. This library is configured to automatically match the closest baseline fluid dynamics model as the starting point for calculations based on the patient's age, gender, weight, and other basic parameters. This method of pre-setting initial values ​​significantly accelerates the convergence speed of the simulation engine and reduces computation time.

[0077] The cloud-based simulation service cluster also includes an automated report generation submodule. It automatically extracts key quantitative indicators from simulation results, such as maximum pressure differential, flow velocity at the narrowest point, and bile stagnation time, and compares them with preset clinical normal ranges. The submodule can automatically generate a preliminary diagnostic report including textual descriptions, key cross-sectional cloud images, and risk assessment conclusions, and send it back to the primary care data collection workstation in encrypted format.

[0078] The collaborative terminal is configured as a lightweight application based on mobile devices. Primary care physicians can directly access cloud-generated diagnostic reports and 3D simulation videos through this terminal. The terminal features a "one-click expert call" function, which automatically pushes the case to the to-do list of relevant experts at higher-level hospitals based on the risk level in the report. On the collaborative terminal of the higher-level experts, the system automatically displays all data reported by the primary care physician and supports experts in quickly annotating and correcting reports via voice messages or gestures.

[0079] In this embodiment, the cloud-based simulation service cluster employs a tiered computing strategy when performing computational tasks. For patients undergoing initial screening, the system performs low-resolution, coarse flow field estimation to quickly identify obvious signs of obstruction; while for complex cases entering the surgical rehearsal stage, the system automatically switches to a high-precision, multi-physics coupled full simulation mode. This differentiated strategy balances computational costs with clinical needs, reducing the overall operating energy consumption of the system.

[0080] The collaborative terminal also includes a built-in doctor-patient communication assistant based on a large language model. It translates complex fluid dynamics parameters and risk assessment conclusions into easily understandable language, assisting primary care physicians in explaining conditions and treatment plans to patients. For example, "risk of inflammatory infiltration due to abnormal local wall shear force" is described as "the bile flow rate is too slow and the pressure is unstable here, easily irritating the vessel wall and causing inflammation and pain." This function improves the efficiency of doctor-patient communication and increases patients' trust and compliance with remote diagnosis and treatment plans.

[0081] The grassroots data acquisition workstation is also equipped with an offline working mode. When the network connection is interrupted, the workstation can temporarily store the acquired images and perform preliminary geometric structure analysis locally, using a pre-built machine learning rule engine to identify suspected critical values. Once the network is restored, the workstation immediately and automatically re-uploads the backlogged data, ensuring the continuity of the cloud simulation service. The cloud simulation service cluster will periodically push the latest software patches and optimized recognition models to the grassroots workstations, achieving synchronous upgrades of system functions.

[0082] The remote collaborative management system for chronic biliary and pancreatic diseases provided in the above embodiments of the present invention, by introducing computational fluid dynamics and digital twin technology, changes the traditional remote image consultation mode that relies on visual experience. The system can not only accurately quantify the physiological functional state of the biliary and pancreatic duct system, but also predict treatment effects through virtual intervention simulation before surgery, reducing the risk of clinical decision-making. The comprehensive application of distributed edge computing, intelligent sensing, and multi-terminal collaboration technologies makes the system suitable not only for high-end research-oriented medical centers, but also effectively empowers a wide range of grassroots medical institutions, promoting the standardization, scientification, and personalization of the diagnosis and treatment of chronic diseases of the biliary and pancreatic system.

[0083] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention. Various changes and improvements can be made to the present invention without departing from its spirit and scope, and all such changes and improvements fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents. Furthermore, the division of functional modules, units, and subunits described in this specification is merely illustrative. In actual engineering implementation, multiple functions can be integrated into the same hardware or software component, or a single function can be divided into multiple more refined physical entities. These do not affect the completeness and effectiveness of the technical solution of the present invention. All numerical values, parameters, and their calculation logic should be reasonably calibrated according to medical engineering practice in the specific application context.

Claims

1. A remote collaborative management system for chronic biliary and pancreatic diseases, characterized in that: include: The medical imaging data acquisition unit is configured to acquire multimodal medical imaging data of patients and extract anatomical structural parameters of the bile and pancreatic duct system. An individualized anatomical model construction unit is connected to the medical imaging data acquisition unit, configured to reconstruct a three-dimensional geometric model of the patient's bile and pancreatic duct system based on the anatomical structural parameters, and given physiological boundary conditions to form an individualized digital twin basic model. The biliary and pancreatic fluid dynamics simulation engine is connected to the individualized anatomical model building unit and is configured to simulate the flow state of biliary and pancreatic fluid under different physiological conditions on the digital twin basic model, and output dynamic flow field parameters including pressure distribution and duct shear force. The virtual intervention simulation unit is connected to the bile and pancreatic fluid dynamics simulation engine and is configured to perform virtual intervention operations based on the dynamic flow field parameters, and to compare and evaluate the changes in fluid dynamic parameters before and after the intervention. The risk warning analysis unit is connected to the virtual intervention simulation unit and is configured to identify hidden obstruction areas based on the dynamic flow field parameters, quantify the degree of dynamic obstacles, and generate graded warning signals. The remote collaborative diagnosis and treatment platform communicates with the above-mentioned units, is configured to integrate simulation analysis results, supports online collaborative consultations among experts from multiple centers, and formulates individualized treatment strategies based on the results of the comparative evaluation.

2. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 1, characterized in that, The medical imaging data acquisition unit includes: The image preprocessing subunit is configured to acquire computed tomography (CT) images and magnetic resonance imaging (MRI) images through a standard medical digital imaging and communication protocol interface, and to perform noise reduction processing on the MRI images using an algorithm based on anisotropic diffusion filtering to eliminate random interference components in the image background. The anatomical feature recognition module is configured to automatically segment the common bile duct, pancreatic duct, gallbladder, and duodenal papilla regions using a deep learning convolutional neural network, determine the spatial topological connections of the bile and pancreatic duct system, and extract the anatomical structural parameters including duct diameter, bending radius, and branch angle. The continuous monitoring module is configured to periodically receive new image data during patient follow-up and trigger the individualized anatomical model construction unit to incrementally correct the existing digital twin basic model. By calculating the change in stenosis rate through image registration technology, it achieves continuous dynamic tracking of disease progression.

3. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 2, characterized in that, The individualized anatomical model construction unit includes: The three-dimensional surface reconstruction submodule is configured to use the moving cube algorithm to convert the anatomical structure parameters into geometric surfaces represented by triangular patches, and to eliminate holes and singular patches on the model surface through a geometric cleanup and optimization algorithm. An adaptive mesh generator is configured to perform fine mesh generation on the narrow, dilated, and branching structures of the bile and pancreatic duct system. It generates unstructured tetrahedral meshes for the anatomical sites at the end of the common bile duct and sets at least five proportionally scaled prism elements on the boundary layer of the duct wall to capture the velocity gradient near the wall. The physical property definition interface is configured to set the physiological boundary conditions, which include a function of the inlet flow rate changing with time, the outlet pressure hydrostatic pressure, and the linear or nonlinear elastic modulus of the pipe wall, set according to the individual patient examination results.

4. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 3, characterized in that, The biliary and pancreatic fluid dynamics simulation engine includes: The governing equation solving module is configured to treat pancreatic and biliary fluid as a non-Newtonian fluid with shear-thinning properties and discretize and solve the continuity equation and Navier-Stokes equation in three-dimensional space. The fluid property adjustment module is configured to set the density of the bile and pancreatic juice as a variable that dynamically changes with the bile concentration, and to set the dynamic viscosity of the bile and pancreatic juice as a function of the shear rate, and to obtain the velocity vector and pressure scalar of each grid cell through iterative calculation. The viscosity correction submodule is configured to receive bile biochemical test indicators and perform nonlinear correction on the characteristic viscosity coefficient of the fluid based on the deviation of cholesterol saturation from the preset benchmark value, so as to reflect the influence of bile stasis on flow resistance. The particle tracking submodule is configured to model the precipitates in bile as discrete phase particles, solve the Langevin equation for the discrete phase particles under the combined effects of fluid resistance, lift, Brownian force and gravity, and predict the probability of particle deposition in pipe walls or narrow passages.

5. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 4, characterized in that, The biliary and pancreatic fluid dynamics simulation engine supports parallel simulation under multiple operating conditions, specifically including: The body position transformation simulation subunit is configured to dynamically adjust the component direction of the gravity vector in the three-dimensional coordinate system according to the different states of the patient, such as supine, lateral, or standing, to simulate the deposition pattern of bile in the gallbladder and the obstruction of its excretion into the common bile duct under different gravity environments. The dietary stimulation simulation subunit is configured to adjust the flow pulse function at the inlet of the common bile duct according to a preset postprandial physiological response model, thereby simulating transient flow field changes during gallbladder contraction and sphincter relaxation. The operating condition comparison module is configured to simultaneously run flow field calculations under fasting, post-meal, and postural changes, and generate a comparative analysis report including a three-dimensional streamline diagram, pressure gradient distribution diagram, time-averaged pipe wall shear force, and local vorticity intensity.

6. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 5, characterized in that, The virtual intervention simulation unit includes: The virtual instrument library module is configured to store various parametric modeling stent models. Each stent model is associated with physical property parameters including stent diameter, length, radial support force, and porosity after expansion. The intervention path planning submodule is configured to determine the location and coverage of stent implantation in virtual space and perform geometric deformation calculations to simulate the topological changes of the original three-dimensional geometric model of the biliary and pancreatic duct system after stent implantation and the mechanical expansion effect on the stenotic site. The comparative evaluation subunit is configured to calculate the pressure loss improvement rate before and after intervention. The pressure loss improvement rate is the percentage obtained by subtracting the pressure difference between the proximal and distal ends before intervention from the pressure difference between the proximal and distal ends after intervention, and then dividing by the pressure difference before intervention. The contact mechanics analysis submodule is configured to calculate the circumferential stress distribution on the bile duct wall after stent release. When the calculated local stress exceeds the elastic limit of the bile duct tissue, it issues a structural damage warning and suggests adjusting the nominal diameter or material stiffness of the virtual stent.

7. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 6, characterized in that, The risk warning and analysis unit includes: The hidden obstruction identification module is configured to use a pressure gradient anomaly detection algorithm to automatically identify a hidden obstruction point when the instantaneous pressure gradient value of a local area exceeds a preset physiological threshold for multiple consecutive peristaltic cycles. The kinetic obstacle quantification module is configured to calculate the bile emptying fraction, which is the ratio of the total amount of fluid passing through the end of the common bile duct to the theoretical amount excreted at the inlet during a complete physiological cycle. The tiered early warning logic module is configured to execute the following logic: When the calculated local maximum shear force exceeds the first preset warning value and the bile emptying fraction is lower than the second preset warning value, a level one high-risk signal is generated. When only a local pressure gradient exceeds the physiological threshold but overall emptying is normal, a secondary medium-risk signal is generated; The feature fusion assessment module is configured to weight and fuse the fluid dynamic parameters with the patient's serum amylase and alkaline phosphatase laboratory indicators, use a Bayesian network to obtain a comprehensive risk score, and trigger the remote collaborative diagnosis and treatment platform to start a multidisciplinary expert consultation process when the comprehensive risk score reaches a predetermined high threshold.

8. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 7, characterized in that, The remote collaborative diagnosis and treatment platform includes: The data integration server is configured to uniformly manage patients' original images, digital twin models, simulation animations, and risk reports, and has a role-based access control model to ensure data security. An interactive visualization interface, built on a 3D graphics rendering engine, is configured to support remote experts in real-time rotation, sectioning, or magnification of the digital twin base model, and synchronously displays real-time changes in pressure or shear force in the form of color contour lines. The virtual endoscopy browsing module is configured to generate simulated endoscopic view images using a camera roaming algorithm, with the internal flow channels of the bile and pancreatic ducts as the path. This allows doctors to perform a panoramic view inside the three-dimensional ductal system to observe the morphology of the lumen in narrowed areas. The collaborative consultation management submodule is configured to provide voice, text, and whiteboard annotation functions, supporting experts to directly perform measurements and annotations on the 3D model and synchronizing the annotation information to all participating terminals in real time; The strategy formulation assistant is configured to automatically rank and recommend the optimal intervention strategy based on the comparison results of multiple schemes output by the virtual intervention simulation unit and using an optimization algorithm.

9. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 8, characterized in that, The system adopts a distributed edge computing architecture, including: An edge computing node deployed within the hospital's local area network includes a local anatomical model generation unit and a distributed dynamics simulation module. It is configured to perform three-dimensional reconstruction using a cluster of graphics processing units and to perform numerical solution tasks of the flow field by storing large sparse matrices in the control equations in a block-circular distribution and using parallel computing capabilities. A central management platform deployed in the cloud includes a virtual intervention global simulator, is configured to maintain a standardized scaffold performance database, and uses an alternative model algorithm to quickly estimate the flow field response after intervention based on lightweight feature data synchronized from edge nodes. A data synchronization agent unit is connected between the edge computing node and the central management platform. It is configured to convert complete grid data into key feature point cloud data and transmit it through an encrypted secure tunnel.

10. The remote collaborative management system for chronic biliary and pancreatic diseases according to claim 9, characterized in that, The system also includes: The automated intervention design module is configured to execute optimization logic based on genetic algorithms. With the objective function of minimizing the total pressure drop in the bile and pancreatic ducts after intervention and homogenizing the local shear force, it automatically searches for the optimal combination of stent diameter, length, and material stiffness within the stent parameter space. The deep risk mining unit is configured to use long short-term memory networks to analyze the coupling relationship between pressure fluctuation cycles and patients' physiological rhythms, identify nonlinear dynamic characteristics, and combine reaction-diffusion equations to simulate the dynamic path of inflammatory mediators in cholestasis with fluid convection and molecular diffusion, predicting the extent of inflammation spread. The blockchain data management module is configured to encapsulate remote consultation opinions, digital twin model versions, and virtual intervention simulation records into timestamped blocks and store them in a distributed manner to ensure the immutability and traceability of the medical decision-making process.

Images