A dynamic medical twin system based on physiological double-helix driving and four-dimensional linkage mode and an interactive prediction method

Abstract

This invention relates to the interdisciplinary field of smart healthcare and digital twin technology, specifically to a dynamic medical twin system and interactive prediction method based on a physiological double helix driven and four-dimensional linkage modalities. The system continuously collects and fuses multi-source biophysical data through a physical entity helix, while a digital virtual helix solves a network of physiological equations coupled with metabolic and stress fields in real time. Both systems achieve endogenous synchronous mapping of gene loci and expressed proteins through a built-in base pairing engine. The four modalities of mirroring, deduction, intervention, and knowledge are responsible for high-fidelity real-time presentation, disease progression prediction, virtual intervention deduction, and clinical knowledge accumulation, respectively, forming a complete closed loop of "observation-deduction-intervention-learning." This invention elevates the system architecture from a static hierarchical stack to a dynamic life-body metaphor, supporting users to perform virtual operations on the twin and receive real-time feedback on multi-scale physiological responses across the entire system, achieving a leap from "morphological simulation" to "life mechanism simulation."

Description

Technical Field

[0001] This invention relates to the field of smart healthcare and digital twin technology, specifically to a dynamic medical twin system and interactive prediction method based on physiological double helix drive and four-dimensional linkage mode. Background Technology

[0002] With the increasing global aging population and the surge in demand for chronic disease management, the traditional medical model is transforming from "passive treatment" to "proactive health management." Digital twin technology has been introduced into the medical field to build virtual patients to assist in clinical decision-making. However, existing medical digital twin or health monitoring systems have four core bottlenecks that make it difficult to meet the needs of accurate prediction and real-time interaction: 1. Static stacking of system architecture, with data and mechanisms separated: The system generally adopts a static layered architecture of "data layer-model layer-application layer". Data acquisition, simulation engine and visualization interface are mechanically stacked, lacking dynamic coupling mechanism. Data flows in one direction and the model cannot adjust the calculation logic autonomously according to real-time data. It can only realize offline simulation or near real-time monitoring and cannot build a living digital twin.

[0003] 2. Low degree of fusion of multi-source heterogeneous data and lack of unified spatiotemporal benchmark: There is a "data silo" problem. Data streams from multiple sources such as millimeter-wave radar, cameras, and environmental sensors are asynchronous and discrete, lacking a sub-millisecond spatiotemporal alignment mechanism. There is a time delay misalignment between the virtual scene and physical reality. In addition, the data acquisition and simulation modules are loosely coupled, and there is a phase difference between their time axes. After alignment, the data cannot drive the model update in real time.

[0004] 3. The prediction model is a black box, lacking interpretability and causal inference ability: It is mostly based on traditional statistics or a single deep learning network, treating patients as static statistical samples and ignoring the dynamic coupling relationship of physiological indicators; the prediction module is mounted as an independent functional layer, with static data snapshots as input and isolated numerical results as output. It cannot interact bidirectionally with the simulation engine and visualization interface, and can only answer "what", but cannot explain "why" or infer "what will happen".

[0005] 4. Limited interaction methods, making it difficult to balance immersion and practicality: The visualization platform is limited to two-dimensional charts or simple three-dimensional model displays, and the front-end interaction and back-end prediction algorithm are not deeply integrated; the interactive interface and core functional modules are connected through multi-level API calls, resulting in significant response delays, making it impossible to achieve synchronous updates of multi-dimensional parameters, and unable to complete the complete operation loop of "observation-deduction-intervention-learning", making it difficult to adapt to the needs of multi-scenario implementation.

[0006] The fundamental flaw of existing technologies lies in the mechanical stacking of architectural paradigms, where functional units are isolated from each other and lack endogenous linkage mechanisms. Therefore, there is an urgent need for a dynamic medical twin system and method that breaks through the traditional stacked architecture and achieves deep integration of data, models, and interactions, so as to promote the application of digital twin technology in personalized medicine. Summary of the Invention

[0007] The purpose of this invention is to propose a dynamic medical twin system and interactive prediction method based on physiological double helix driving and four-dimensional linkage mode, so as to realize the real-time solution and interactive intervention deduction of physiological equation network under multi-physics field, and improve the interpretability of disease prediction and the foresight of clinical decision-making.

[0008] According to a first aspect of the embodiments of this disclosure, a dynamic medical twin system based on physiological double helix drive and four-dimensional linkage modality is provided, comprising: The physical entity spiral module integrates multi-source heterogeneous biophysical data in real time to generate and continuously update the state gene spectrum that characterizes the real-time physiological state of patients. The digital virtual spiral module dynamically interacts with the physical entity spiral module and has a built-in editable physiological equation network. The physiological equation network is used to solve the metabolic flow state and stress field in real time based on the state gene spectrum, and to generate and continuously update the trend gene spectrum that characterizes the future disease trend of the patient. The four-dimensional linkage modal module includes a mirror modal, a deduction modal, an intervention modal, and a knowledge modal, which are used to faithfully present and interact with the current state, future trends, virtual intervention results, and derived clinical knowledge of the digital virtual spiral module. The base pairing engine defines the core parameters in the physical entity helical module and the digital virtual helical module as gene loci, defines the functional expression in the four-dimensional linkage modal module as expressed proteins, and establishes an endogenous, one-to-one synchronous mapping relationship between the gene loci and the expressed proteins based on a preset physiological rule base; when any gene locus changes, the base pairing engine drives all associated expressed proteins to update synchronously.

[0009] In one embodiment, the physical entity spiral module includes: Multi-source asynchronous sensing unit is used to connect to wearable devices, implantable sensors, environmental sensing nodes and medical imaging equipment to acquire asynchronous, multimodal raw biophysical data streams; The spatiotemporal alignment and feature extraction unit is used to apply sub-millisecond-level timestamp alignment to the original biophysical data stream and to perform cross-modal feature extraction without data delocalization through an embedded federated learning node to generate the state gene spectrum.

[0010] In one embodiment, the digital virtual spiral module includes a metabolism-stress field coupled solver, used to express the life processes of blood circulation, tissue deformation, cell metabolism and electrical conduction as a set of coupled partial differential equations; wherein, the metabolism-stress field coupled solver uses the stress distribution calculation results of the mechanical field as the input parameters of the metabolic field equation, and uses the material exchange results of the metabolic field as the boundary conditions of the mechanical field, to realize the dynamic linkage and real-time solution of the two.

[0011] In one embodiment, the mirror mode includes: The object field probe tool is used to respond to the user's point selection operation at any location on the 3D twin, retrieve and display the multiphysics field data of that point in real time, and automatically plot the curve of the data changing over time. The cross-scale spatiotemporal adjustment unit allows users to continuously adjust the time scale and spatial area of ​​interest within the same view, enabling cross-scale visualization from microscopic cellular metabolism to macroscopic tissue deformation.

[0012] In one embodiment, the inference mode includes: A hybrid prediction model constrained by physical laws is proposed. The model adopts a hybrid architecture of neural differential equations and graph attention mechanism, and embeds the physical laws represented by the physiological equation network as strong constraints to generate disease course prediction results with physiological interpretability. The differential staining visualization unit is used to overlay the future state corresponding to the prediction result with the current state in three-dimensional space, and to intuitively present the spatial range and intensity of the disease evolution through gradient staining.

[0013] In one embodiment, the intervention modality allows the user to directly apply virtual intervention commands to the three-dimensional twin presented by the mirror modality or the deductive modality; the base pairing engine responds to the virtual intervention commands by converting them into parameter adjustments to the physiological equation network in the digital virtual spiral module, driving the digital virtual spiral module to complete the re-solution and real-time feedback of the multi-scale physiological response of the entire system within a preset time threshold.

[0014] In one embodiment, the knowledge modality includes: An interpretable report generator is used to automatically associate and output a corresponding underlying physiological mechanism interpretation report for each risk warning or prediction result generated for the inference modality. An evolutionary case knowledge base is used to structure and store interactive data, decision paths, and results generated throughout the entire process of mirroring, deduction, and intervention, forming a dynamic knowledge graph that can be queried and mined.

[0015] According to a second aspect of the present disclosure, a dynamic medical twin interaction prediction method applied to the above-described system is provided, comprising: The system uses a physical entity spiral module to fuse multi-source biophysical data in real time, generating and continuously updating the patient's state gene spectrum. The physiological equation network built into the digital virtual spiral module is used to solve the metabolic flow state and stress field in real time based on the state gene spectrum, thereby generating and continuously updating the trend gene spectrum. Through the base pairing engine, the changes in gene loci in the state gene spectrum and trend gene spectrum are mapped in real time to the mirror, inference, intervention and knowledge modalities of the four-dimensional linkage modal module, driving the synchronous update of the functional expression of each modality. The system presents the patient's current physiological state in the mirror mode, generates interpretable disease course predictions in the extrapolation mode, receives virtual intervention instructions and provides real-time feedback on the entire system's response in the intervention mode, and accumulates clinical knowledge and decision-making basis in the knowledge mode.

[0016] In one embodiment, the real-time bidirectional coupling solution of metabolic flow and stress field is as follows: the stress distribution calculation result of the mechanical field is used as the input parameter of the Monod equation of the metabolic field to obtain the local adenosine triphosphate production rate and lactic acid accumulation trend in real time; and the material exchange result of the metabolic field is used as the boundary condition of the mechanical field to adjust the tissue deformation and stress distribution in real time.

[0017] In one embodiment, when the physical entity spiral module detects an abnormal change in the gene site of vascular wall stress, the base complementarity pairing engine synchronously triggers the re-solution of the corresponding site of the myocardial metabolism equation in the digital virtual entity spiral module according to the AT / CG pairing rule. At the same time, it drives the mirror mode to change the stress color of the corresponding region, update the risk curve of the extrapolation mode, highlight the recommended options of the intervention mode, and generate temporary warning records of the knowledge mode.

[0018] The advantages of the above technical solutions adopted in this invention compared with the prior art are as follows: This invention enables the construction of a unique, computable, and interactive personalized digital life form for each patient, breaking through the limitations of traditional medical digital twin technology. The physical entity spiral module achieves precise fusion and feature extraction of multi-source heterogeneous data, while the digital virtual entity spiral module completes the dynamic linkage solution of metabolic flow and stress fields. The double helix, combined with a base complementarity pairing engine, achieves real-time synchronous mapping of parameters and functions across modules. The four-dimensional linkage modality enables high-fidelity cross-scale presentation of the patient's physiological state, physiologically interpretable disease progression prediction, and supports immediate feedback of virtual intervention commands, while simultaneously accumulating clinical knowledge to form a dynamic knowledge graph.

[0019] This invention upgrades digital twins from morphological simulation to life mechanism simulation, providing a new technological carrier for medical treatment to delve deeper from genotype to phenotypic dynamics. It also creates innovative human-computer interaction interfaces for scenarios such as medical education and remote collaborative diagnosis and treatment, greatly improving the foresight and scientific nature of clinical decision-making and adapting to the needs of medical and health management in multiple scenarios. Attached Figure Description

[0020] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an undue limitation of this application.

[0021] Figure 1 This is an architecture diagram of a dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode; Figure 2 This is a flowchart of a dynamic medical twin interaction prediction method based on physiological double helix driving and four-dimensional linkage modality. Detailed Implementation

[0022] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of this disclosure. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using a dedicated hardware-based system that performs the specified functions or operations, or using a combination of dedicated hardware and computer instructions.

[0026] Example 1: like Figure 1 As shown, this embodiment discloses a dynamic medical twin system based on physiological double helix drive and four-dimensional linkage modality. This system abandons the static hierarchical architecture of traditional medical digital twins, constructing an organic, life-like holistic architecture. It achieves a leap from morphological simulation to life mechanism simulation, creating a unique, computable, and interactive personalized digital life form for each patient. It adapts to various medical needs such as personalized pathology deduction, clinical intervention simulation, medical education, and remote diagnosis and treatment. Specifically, it includes a physical entity helical module, a digital virtual entity helical module, a four-dimensional linkage modal module, and a base complementarity pairing engine. Each module achieves deep integration of data, models, and interactions through an endogenous linkage mechanism, as detailed below: The physical entity spiral module is used to fuse multi-source heterogeneous biophysical data in real time to generate and continuously update the state gene spectrum representing the real-time physiological state of patients. This module includes a multi-source asynchronous sensing unit and a spatiotemporal alignment and feature extraction unit. The former connects to wearable devices, implantable sensors, environmental sensing nodes, and medical imaging equipment to acquire asynchronous, multimodal raw biophysical data streams. The latter applies sub-millisecond-level timestamp alignment to the raw biophysical data streams and completes cross-modal feature extraction without data de-localization through implanted federated learning nodes, achieving accurate fusion of multi-source data in the spatiotemporal dimensions.

[0027] The digital virtual spiral module dynamically feeds back to the physical solid spiral module. It has a built-in editable physiological equation network for real-time bidirectional coupling solution of metabolic flow and stress field based on the state gene spectrum, generating and continuously updating the trend gene spectrum characterizing the future disease course of patients. The module has a built-in metabolic-stress field coupling solver, which expresses life processes such as blood circulation and tissue deformation as a set of mutually coupled partial differential equations. The stress distribution calculation results of the mechanical field are used as input parameters of the metabolic field equation, and the material exchange results of the metabolic field are used as boundary conditions of the mechanical field. The two are dynamically linked and solved in real time based on GPU parallelism.

[0028] The four-dimensional linkage modal module, serving as the presentation and interaction interface for the double helix driving force, comprises four sub-modalities: mirror modality, deductive modality, intervention modality, and knowledge modality. Among them, the mirror modality, through the object field probe tool and the cross-scale spatiotemporal adjustment unit, achieves a high-fidelity real-time presentation of the patient's physiological state across scales. The deductive modality, relying on a hybrid prediction model constrained by physical laws and a differential staining visualization unit, generates physiologically interpretable disease course predictions and intuitively presents the trend of disease progression. The intervention modality allows users to apply virtual intervention commands, driving the digital virtual helix module to complete real-time feedback of multi-scale physiological responses across the entire system. The knowledge modality, through an interpretable report generator and an evolutionary case knowledge base, completes the interpretation of physiological mechanisms and the accumulation of clinical knowledge, forming a dynamic knowledge graph.

[0029] The base complementarity pairing engine is the core connection mechanism of the system. It defines the core parameters of the double helix module as gene sites and the functional expression of the four-dimensional linkage modality module as expressed proteins. Based on a preset physiological rule library, it establishes an endogenous and one-to-one synchronous mapping relationship between the two. When any gene site changes, the engine automatically and instantaneously drives all related expressed proteins to update synchronously according to the AT / CG pairing rule, completely eliminating the time lag and misalignment of data between functional units, and ensuring the high fidelity, real-time performance and interpretability of the system.

[0030] Simultaneously, a comprehensive trusted data acquisition system is constructed, encompassing multi-source asynchronous sensing, dynamic quality assessment, and compliance and privacy enhancement. This system deeply adapts technologies such as federated learning and blockchain notarization to medical scenarios, enabling real-time feature extraction and quality verification without data leaving the domain. This provides the entire system with a real-time, legal, and high-fidelity multi-physics data source.

[0031] The above modules can be deployed on the same device or distributed devices; the division of modules is only a functional logic description and does not limit the specific physical boundaries or implementation order.

[0032] The system automatically synchronizes the latest physiological monitoring data of patients every 24 hours, completes the spatiotemporal alignment and feature fusion of multi-source data, updates the "state gene spectrum" of the physical entity spiral, and simultaneously triggers a full re-solution of the digital virtual spiral, ensuring that the digital twin maintains a real-time and accurate match with the patient's actual physiological state. Based on continuously added real-world clinical data, the system's built-in neural differential equation-graph attention hybrid model will automatically conduct incremental iterative training, continuously optimize the model's predictive performance, and improve the accuracy of long-term disease prediction. In this embodiment, the accuracy of atrial fibrillation risk prediction at 6 months has increased from 85% to 92%. The knowledge modality regularly performs in-depth mining and analysis on the case data accumulated throughout the entire process, mining and refining new pathological association patterns (such as specific genotypes and drug response patterns), and updates the mining results to the system's physiological rule base in the form of a knowledge graph, providing more accurate mapping rules for the base complementarity pairing engine. The system supports open-source extensions of physiological equation networks. Users can upload coupling equation templates adapted to different diseases such as diabetes and kidney disease, as well as different physiological systems such as the respiratory system and the nervous system, to continuously expand the simulation coverage of the system and gradually realize the functional upgrade from single-organ twins to whole-body multi-organ twins.

[0033] Example 2: like Figure 2 As shown, this embodiment discloses a dynamic medical twin interaction prediction method applied to the system described in Embodiment 1. This method relies on a double-helix driven architecture and a four-dimensional linkage modality to achieve a closed-loop process encompassing real-time mapping of patient physiological states, accurate prediction of disease progression trends, immediate feedback from virtual intervention, and accumulation of clinical knowledge. This significantly improves the foresight and scientific rigor of clinical decision-making, providing technical support for precision medicine to move from genotype to phenotypic dynamics. Specifically, it includes the following steps: Step 1: Collect and fuse multi-source biophysical data through the physical entity spiral module to generate and continuously update the patient's state gene spectrum; specifically, the original biophysical data streams from multiple data sources such as wearable devices and medical imaging equipment are accessed through the multi-source asynchronous sensing unit, and then the spatiotemporal alignment and feature extraction unit completes sub-millisecond timestamp alignment and cross-modal feature extraction without data delocalization, so as to realize the spatiotemporal accurate fusion of multi-source heterogeneous data and continuously update the state gene spectrum that represents the patient's real-time physiological state.

[0034] Step 2: Using the physiological equation network built into the digital virtual spiral module, the metabolic flow and stress field are solved in real time through bidirectional coupling based on the state gene spectrum, generating and continuously updating the trend gene spectrum. Specifically, the stress distribution calculation results of the mechanical field are used as input parameters of the metabolic field Monod equation through the metabolic-stress field coupled solver to calculate the local adenosine triphosphate production rate and lactate accumulation trend in real time. At the same time, the material exchange results of the metabolic field are used as the boundary conditions of the mechanical field to adjust tissue deformation and stress distribution in real time, realizing the dynamic linkage and real-time solution of the two, and continuously updating the trend gene spectrum that characterizes the future disease trend of the patient.

[0035] Step 3: Through the base complementation engine, the changes in gene loci in the state gene spectrum and trend gene spectrum are mapped in real time to each sub-modality of the four-dimensional linkage modality module, driving the synchronous update of the functional expression of each modality; when the physical entity spiral module detects abnormal changes in gene loci such as vascular wall stress, the engine triggers the re-solution of the corresponding physiological equation loci in the digital virtual spiral module according to the AT / CG pairing rule, and drives the mirror modality, inference modality, intervention modality and knowledge modality to complete the synchronous update, ensuring seamless linkage of each module.

[0036] Step 4: Through the four-dimensional linkage modal module, multi-dimensional collaborative interaction and clinical knowledge accumulation are completed. In the mirror modality, the current physiological state of the patient is presented in high-fidelity real-time through the object field probe tool and the cross-scale spatiotemporal adjustment unit. In the deduction modality, the disease course prediction results with physiological interpretability are generated through the hybrid prediction model constrained by physical laws, and the disease progression trend is presented intuitively through the differential staining visualization unit. In the intervention modality, the virtual intervention command applied by the user is received, and the digital virtual spiral module is driven to complete the re-solution and real-time feedback of the multi-scale physiological response of the whole system within the preset time threshold. In the knowledge modality, the underlying physiological mechanism interpretation report is generated for each risk warning, and the interaction data, decision path and results of the whole process are structured and stored, which are accumulated into queryable and mineable clinical knowledge and form a dynamic knowledge graph that is constantly evolving.

[0037] This method, through the steps described above, achieves a closed-loop process from patient physiological data collection to clinical knowledge accumulation. It supports causal interaction between users and patients' personalized digital life forms, shifting clinical intervention from post-hoc statistics to pre-hoc virtual experiments, effectively reducing diagnostic and treatment risks. At the same time, the accumulated clinical knowledge can feed back into medical teaching and research, possessing good industrialization and promotion value.

[0038] This embodiment focuses on a 55-year-old male patient with hypertension and left ventricular hypertrophy. The patient has had hypertension for 8 years, takes 5mg of antihypertensive drug A orally daily, and has no comorbidities such as diabetes or coronary heart disease. He is 175cm tall, weighs 80kg, has a resting blood pressure of 165 / 105mmHg, and a left ventricular wall thickness of 14mm. Figure 2 The method flowchart is executed using a distributed software architecture that integrates edge and cloud. The front end is built on the UnityHDRP engine to create an interactive 3D visualization interface, while the back end deploys a dual-helix driving core service and a four-modal linkage engine. Each module achieves data communication and logical linkage through a base complementation engine. Specifically, the edge is deployed on patient home terminal tablets and wearable devices with corresponding edge gateways, responsible for data acquisition and preprocessing, sub-millisecond timestamp initial alignment, and loading of simplified health views, supporting offline data caching and low-latency interaction. The cloud core service is deployed on the doctor's workstation GPU server, equipped with real-time solvers for physical and digital virtual helices, a base complementation engine, and four-modal inference and knowledge generation modules, providing core computing power and algorithmic support for the system. Multi-role interactive terminals are deployed on the doctor's workstation, patient home tablets, and medical teaching VR devices, respectively loading clinical diagnosis and treatment views, family health views, and teaching and training views, achieving multi-scenario adaptation of a single system.

[0039] Patient digital profile construction: Administrators log in through the system platform in administrator view to complete the standardized entry of basic patient information, medical history, medication history, and clinical diagnosis results, and simultaneously import outpatient examination data such as cardiac MRI images, Holter ECG, and ultrasound elastography from the past 18 months. The system automatically performs data format verification and desensitization processing, and generates a unique digital twin identifier for each patient.

[0040] Physical entity spiral initialization: Based on imported 7T cardiac MRI image data, the system automatically completes anatomical structure segmentation and 3D reconstruction, generating a 1:1 scale digital twin of the heart. It automatically identifies key anatomical structures such as the left ventricle and coronary arteries, and extracts initial anatomical parameters such as left ventricular wall thickness and left anterior descending coronary artery diameter. Simultaneously, the system connects to wearable devices (smart bracelets, blood pressure monitors), implantable hemodynamic sensors, and environmental sensing nodes (temperature, humidity, air quality), establishing a real-time data stream channel and configuring sub-millisecond spatiotemporal alignment rules. This unifies discrete medical images, continuous physiological signals, and environmental readings onto the same timeline, forming an initial version of the patient's personalized "state gene spectrum."

[0041] Digital Virtual Body Spiral Initialization: Through the system's equation editing interface, users can call upon the built-in physiological equation network template. Targeting the pathological characteristics of hypertension complicated by left ventricular hypertrophy, the system integrates four solvers: hemodynamics, tissue mechanics, cell metabolism, and myocardial electrical conduction. The system automatically substitutes the patient's anatomical parameters and historical treatment data into the equation set, completing the initial configuration of personalized parameters such as blood viscosity adjustment factor, myocardial elastic modulus, and cellular oxygen extraction rate. This generates a patient-specific, editable physiological equation network, i.e., the initial version of the "Trend Gene Profile."

[0042] Base Complementary Pairing Engine Configuration: The system automatically establishes a mapping relationship between gene loci and expressed proteins based on a preset physiological rule base (AT / CG pairing rules). For example, it pairs vascular wall stress sites in the physical helix with myocardial metabolic equation sites in the digital virtual helix, and simultaneously maps stress value changes to the color gradient of the mirror mode, the risk curve of the extrapolated mode, the recommended options of the intervention mode, and the warning records of the knowledge mode. The pairing engine completes initialization and debugging to ensure that the simulation calculation latency is ≤50ms and the data synchronization accuracy is ≤0.1ms.

[0043] Four-dimensional linkage modal activation: The system automatically generates a three-dimensional high-fidelity cardiac twin and activates the real-time rendering pipeline of the mirror modality; the derivation modality loads a pre-trained neural differential equation-graph attention hybrid model; the intervention modality opens a virtual operation interface; and the knowledge modality initializes the case knowledge base. The system enters standby mode, awaiting user interaction.

[0044] Users can access the system's data management interface to view the real-time status of multi-source data collected by the physical helical spiral, including wearable physiological signals, implantable hemodynamic data, environmental sensing data, and clinical imaging data. The system automatically performs sub-millisecond timestamp alignment, federated learning feature extraction, homomorphic encrypted transmission, and blockchain notarization, generating a data fusion quality report that displays the synchronization accuracy (≤0.1ms), feature extraction results, and data compliance notarization hash values ​​of each data source. The physical helical spiral continuously updates the "state gene spectrum" and feeds it to the digital virtual helical spiral in real time. Based on the latest "state gene," the digital virtual helical spiral calls the GPU parallel solver to resolve the metabolic-stress field coupling equations within 42ms, updating the "trend gene spectrum." The two spirals achieve dynamic mutual feedback through a bidirectional data channel. For example, when the physical helical spiral detects a sudden increase in the patient's blood pressure, the digital virtual helical spiral immediately recalculates the left ventricular wall stress distribution and feeds the result back to the physical helical spiral, triggering a priority adjustment for the next round of data acquisition.

[0045] Users can access the system's 3D twin interactive interface and perform basic operations such as rotation, scaling, and translation on the 3D twin of the heart. Its core innovative functions include: Spatial area delineation: Users can select areas on the twin by dragging a box with the mouse, with dimensions of 0.1mm. 3 With precise definition of the area of ​​interest, the system automatically highlights the anatomical structure of that area and simultaneously displays the multi-physics correlation data (stress, blood flow velocity, oxygen partial pressure, pH value) of that area.

[0046] Timescale adjustment: Users can freely adjust the observation timescale by dragging the time slider at the bottom of the interface, from a millisecond-level scale of 0.1ms / frame (such as action potential propagation) to a monthly-level scale of 1 month / frame (such as myocardial remodeling), achieving observations across 10 timescales.9 Continuous observations on a time scale of magnitude.

[0047] Field Probe Query: Users can activate the field probe tool by clicking on any location of the twin. The interface will pop up a real-time data panel, displaying multiple physical field parameters such as stress value, shear force, oxygen partial pressure, and pH value at that point, and automatically plotting the data change curves of that point over the past 24 hours and the past 18 months.

[0048] Differential staining viewing: Users click on the differential staining mode, select the time point for comparison (such as the current state versus the state without intervention for 6 months), and the system automatically generates a three-dimensional overlay model. The spatial range and intensity of lesion evolution are intuitively presented through red-yellow-green gradient staining, and high-risk areas (stress ≥14kPa) are automatically marked.

[0049] Users enter the intervention simulation interface, select the prediction duration (e.g., 3 months, 6 months, 1 year), and click "Start Prediction." The system invokes a neural differential equation-graph attention hybrid model embedded with strong constraints of physiological laws, combining real-time simulation data with historical trend data to output a prediction report within 3 seconds. The report includes the changing trends of core physiological parameters, high-risk areas of lesions, and the probability of complication. All prediction results are accompanied by interpretations of physiological mechanisms. For example, the system can explain that "increased stress on the anterior wall of the left ventricle is due to poor blood pressure control leading to increased vascular wall load, which in turn causes abnormal myocardial cell metabolism and myocardial fibrosis." The simulation results are visualized on a 3D twin using differential staining, allowing users to intuitively see the spatial distribution of future lesion areas.

[0050] In the intervention simulation interface, users can freely set intervention plans through a visual operation module, including adjusting drug dosage and frequency, simulating surgical interventions (such as stent implantation or ablation), and adjusting environmental temperature and humidity. After setting, clicking "Execute Simulation" automatically transforms the intervention plan into parameter and boundary condition adjustments for a physiological equation network. This is fed back in real-time to the double-helix driving force via a base pairing engine, completing a full system re-simulation within 500ms. The system outputs post-intervention 3D twin state changes, comparisons of core physiological parameters, changes in complication risk, and generates an intervention plan effectiveness evaluation report. For example, a doctor can increase the dosage of antihypertensive drug A from 5mg to 6mg and add 10mg of vasodilator B. The system will immediately provide feedback on changes in left ventricular stress distribution, lactate accumulation trends, and atrial fibrillation risk probability, helping doctors optimize treatment plans.

[0051] The system automatically records data from every mirror observation, simulation prediction, and intervention operation, including user operation logs, simulation parameter changes, output results, and decision-making paths. For each risk warning and prediction result, the system automatically generates a corresponding physiological mechanism interpretation report and stores it in a case knowledge base. The knowledge base uses a graph database structure, linking individual patient data with group knowledge, and supports doctors to query via natural language (e.g., "comparison of intervention effects in similar cases"). Simultaneously, the system can periodically generate knowledge briefs summarizing the evolution of patient conditions and intervention response characteristics, providing data support for clinical research.

[0052] When the physical entity spiral detects a 5% increase in vascular wall stress at a site (analogous base A), the pairing engine automatically triggers a re-solution of the corresponding myocardial metabolism equation site (analogous base T) in the digital virtual spiral based on a preset rule base (AT pairing). Simultaneously, the pairing engine maps this change to four modalities: the mirror modal changes the stress color of the region from yellow to red; the extrapolation modal automatically updates the risk curve for future myocardial hypertrophy; the intervention modal highlights recommended drugs that may reduce stress in the medication adjustment options; and the knowledge modal generates a temporary record: "Persistently high vascular wall stress indicates a risk of myocardial compensation; attention to XX is recommended." The entire process is completed within 100ms, with all modalities updated synchronously, without any data delay or interface lag.

[0053] Application Example 1: Home-based chronic disease management for hypertensive patients Mr. Zhang, a 58-year-old patient, had a 10-year history of hypertension and regularly took antihypertensive medication, but his blood pressure control was ineffective. The system collected multi-source data in real time through a wearable wristband, a smart blood pressure monitor, and a home environment sensing node. The physical spiral module continuously updated the patient's personalized "status gene spectrum." One day, the system detected persistently high nocturnal blood pressure. The digital virtual spiral module, through analysis, concluded that left ventricular wall stress was increasing at a rate of 0.4 kPa per month, and the risk of myocardial hypertrophy within the next 6 months was 75%. The mirror modality immediately marked the high-risk area in red and pushed a risk warning through the family health view. The patient adjusted his medication virtually through the intervention modality, increasing his antihypertensive medication dosage from 5 mg to 6 mg. The system immediately reported a 12% decrease in left ventricular stress and a reduction in the risk of atrial fibrillation from 8% to 4% under this plan. After submitting the optimized plan to his attending physician for confirmation, the patient officially adjusted his medication. The system continuously tracked the medication effect, and a follow-up examination three months later showed that the rate of increase in left ventricular wall thickness had decreased from 0.8 mm / year to 0.3 mm / year. Knowledge modality automatically generates chronic disease management reports, summarizing patients' response characteristics to drug dosage adjustments, and providing data reference for the subsequent development of personalized treatment plans.

[0054] Application Example 2: Postoperative Rehabilitation Monitoring for Coronary Artery Disease Ms. Li, a 65-year-old patient, was enrolled in the postoperative rehabilitation monitoring system one month after coronary artery stent implantation. The system uses an implanted hemodynamic sensor to monitor key indicators such as blood flow velocity and vessel wall stress in the stent segment in real time. The physical spiral module simultaneously integrates multi-source data including electrocardiogram, blood pressure, and daily activity levels, dynamically updating the "state gene spectrum." One day, the system detected an abnormally slowed blood flow velocity distal to the stent. The digital virtual spiral module showed a 60% probability of local thrombosis. The mirror modality accurately marked the stenotic area on a three-dimensional twin of the heart and displayed a key data point—a 35% decrease in shear force in that area—using a field probe tool. The doctor increased the antiplatelet drug dosage through intervention modality simulation. The system immediately reported that the thrombosis risk decreased to 15% under this plan, but the bleeding risk slightly increased. After comprehensive evaluation, the doctor decided to maintain the original medication plan while also increasing the patient's daily walking volume to improve local blood flow, based on the system's recommendations. The system generates personalized rehabilitation reports daily to guide the patient in scientifically controlling exercise intensity. The knowledge modality provides a reference for doctors to dynamically adjust rehabilitation plans by comparing the rehabilitation trajectories of similar postoperative cases. Six months later, the patient underwent a follow-up examination, and the blood flow in the stent segment was unobstructed, with the recovery effect meeting expectations.

[0055] Application Example 3: Medical Teaching and Clinical Training A cardiology department at a medical school applied this system to clinical teaching and training. Instructors imported desensitized digital twins of real patients into the system's teaching view, allowing students to immerse themselves in a virtual cardiac scenario using VR devices for practical training. Students could drag a timeline to observe the entire eight-year progression of a patient's disease, from early hypertension to left ventricular hypertrophy and then to compensated heart failure. At any point in the disease progression, students could activate the equation visualization function to view the coupling equations and specific parameter values ​​of the dominant physiological processes, intuitively understanding the dynamic coupling relationship between hemodynamics and cellular metabolism. Simultaneously, students could independently perform virtual interventions, such as trying different combinations of antihypertensive drugs and adjusting dosages, observing the real-time impact of these actions on ventricular remodeling. The system automatically recorded each student's operational path and intervention results, generating personalized reports based on the knowledge modality, comparing student decisions with standard clinical protocols, and providing detailed explanations of the physiological mechanisms. This teaching model effectively improved students' understanding and application of complex cardiovascular pathophysiology, increasing training effectiveness by 40% compared to traditional teaching methods.

[0056] Those skilled in the art will understand that the modules or steps described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, which can then be stored in a storage device for execution by a computer device. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. This disclosure is not limited to any particular combination of hardware and software.

[0057] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0058] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode, characterized in that, include: The physical entity spiral module integrates multi-source heterogeneous biophysical data in real time to generate and continuously update the state gene spectrum that characterizes the real-time physiological state of patients. The digital virtual spiral module dynamically interacts with the physical entity spiral module and has a built-in editable physiological equation network. The physiological equation network is used to solve the metabolic flow state and stress field in real time based on the state gene spectrum, and to generate and continuously update the trend gene spectrum that characterizes the future disease trend of the patient. The four-dimensional linkage modal module includes a mirror modal, a deduction modal, an intervention modal, and a knowledge modal, which are used to faithfully present and interact with the current state, future trends, virtual intervention results, and derived clinical knowledge of the digital virtual spiral module. The base pairing engine defines the core parameters in the physical entity helical module and the digital virtual helical module as gene loci, defines the functional expression in the four-dimensional linkage modal module as expressed proteins, and establishes an endogenous, one-to-one synchronous mapping relationship between the gene loci and the expressed proteins based on a preset physiological rule base; when any gene locus changes, the base pairing engine drives all associated expressed proteins to update synchronously.

2. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The physical entity spiral module includes: Multi-source asynchronous sensing unit is used to connect to wearable devices, implantable sensors, environmental sensing nodes and medical imaging equipment to acquire asynchronous, multimodal raw biophysical data streams; The spatiotemporal alignment and feature extraction unit is used to apply sub-millisecond-level timestamp alignment to the original biophysical data stream and to perform cross-modal feature extraction without data delocalization through an embedded federated learning node to generate the state gene spectrum.

3. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The digital virtual spiral module includes a metabolism-stress field coupled solver, which is used to express the life processes of blood circulation, tissue deformation, cell metabolism and electrical conduction as a set of coupled partial differential equations. The metabolism-stress field coupled solver uses the stress distribution calculation results of the mechanical field as the input parameters of the metabolic field equations and the material exchange results of the metabolic field as the boundary conditions of the mechanical field, so as to realize the dynamic linkage and real-time solution of the two.

4. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The mirror mode includes: The object field probe tool is used to respond to the user's point selection operation at any location on the 3D twin, retrieve and display the multiphysics field data of that point in real time, and automatically plot the curve of the data changing over time. The cross-scale spatiotemporal adjustment unit allows users to continuously adjust the time scale and spatial area of ​​interest within the same view, enabling cross-scale visualization from microscopic cellular metabolism to macroscopic tissue deformation.

5. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The inference modes include: A hybrid prediction model constrained by physical laws is proposed. The model adopts a hybrid architecture of neural differential equations and graph attention mechanism, and embeds the physical laws represented by the physiological equation network as strong constraints to generate disease course prediction results with physiological interpretability. The differential staining visualization unit is used to overlay the future state corresponding to the prediction result with the current state in three-dimensional space, and to intuitively present the spatial range and intensity of the disease evolution through gradient staining.

6. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The intervention modality allows users to directly apply virtual intervention commands to the three-dimensional twin presented in the mirror or deductive modality; the base pairing engine responds to the virtual intervention commands by converting them into parameter adjustments to the physiological equation network in the digital virtual spiral module, driving the digital virtual spiral module to complete the re-solution and real-time feedback of the multi-scale physiological response of the entire system within a preset time threshold.

7. The dynamic medical twin system based on physiological double helix drive and four-dimensional linkage mode as described in claim 1, characterized in that, The knowledge modalities include: An interpretable report generator is used to automatically associate and output a corresponding underlying physiological mechanism interpretation report for each risk warning or prediction result generated for the inference modality. An evolutionary case knowledge base is used to structure and store interactive data, decision paths, and results generated throughout the entire process of mirroring, deduction, and intervention, forming a dynamic knowledge graph that can be queried and mined.

8. A dynamic medical twin interaction prediction method applied to the above system, characterized in that, include: The system uses a physical entity spiral module to fuse multi-source biophysical data in real time, generating and continuously updating the patient's state gene spectrum. The physiological equation network built into the digital virtual spiral module is used to solve the metabolic flow state and stress field in real time based on the state gene spectrum, thereby generating and continuously updating the trend gene spectrum. Through the base pairing engine, the changes in gene loci in the state gene spectrum and trend gene spectrum are mapped in real time to the mirror, inference, intervention and knowledge modalities of the four-dimensional linkage modal module, driving the synchronous update of the functional expression of each modality. The system presents the patient's current physiological state in the mirror mode, generates interpretable disease course predictions in the extrapolation mode, receives virtual intervention instructions and provides real-time feedback on the entire system's response in the intervention mode, and accumulates clinical knowledge and decision-making basis in the knowledge mode.

9. The dynamic medical twin interaction prediction method according to claim 8, characterized in that, The real-time bidirectional coupling solution of metabolic flow and stress field is as follows: the stress distribution calculation results of the mechanical field are used as the input parameters of the Monod equation of the metabolic field to obtain the local adenosine triphosphate production rate and lactic acid accumulation trend in real time. The results of material exchange in the metabolic field are used as the boundary conditions of the mechanical field to adjust tissue deformation and stress distribution in real time.

10. The dynamic medical twin interaction prediction method according to claim 8, characterized in that, When the physical entity spiral module detects an abnormal change in the gene locus of vascular wall stress, the base complementarity pairing engine, according to the AT / CG pairing rule, synchronously triggers the re-solution of the corresponding site of the myocardial metabolism equation in the digital virtual entity spiral module. At the same time, it drives the mirror mode to change the stress color of the corresponding region, update the risk curve of the extrapolation mode, highlight the recommended options of the intervention mode, and generate temporary warning records of the knowledge mode.

Images