Digital twin-based drug delivery and dosage assessment system and its application

By using a digital twin model of pediatric physiological pharmacokinetics, combined with sensor data and neural network technology, the problem of medication error caused by physiological parameter variability has been solved, achieving precise control and safety assurance of pediatric medication.

CN115458175BActive Publication Date: 2026-11-17BEIJING GENERAL AEROSPACE HOSPITAL
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202211212465.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-09-29
Publication Date
2026-11-17
Estimated Expiration
2042-09-29

AI Technical Summary

Technical Problem

Existing physiological pharmacokinetic models have significant errors in extrapolating from adults to children and cannot effectively address the variability of physiological parameters, making it difficult to determine drug dosages for children.

Method used

A digital twin model of physiological pharmacokinetics is adopted. By establishing a digital twin model of physiological pharmacokinetics in children, and combining sensor data, virtual entity modules and twin data modules, personalized drug dosage assessment can be achieved, and dosage adjustment can be carried out using neural networks and dynamic data-driven analysis technology.

Benefits of technology

It enables precise control and safety assurance of medication dosage for children, and can update personalized models based on individual growth and development characteristics, reducing prediction errors and improving the accuracy of medication effect judgment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115458175B_ABST
    Figure CN115458175B_ABST
Patent Text Reader

Abstract

This invention provides a drug administration and dosage assessment system based on digital twins and its application. The drug administration and dosage assessment system includes a module for inputting basic information of the drug recipient, a module for acquiring physical entity data of physiological pharmacokinetics, a virtual entity module of physiological pharmacokinetics, a twin data module, and an output module. The drug administration and dosage assessment system of this invention can first simulate the drug administration on the physiological pharmacokinetics virtual entity module before administering the drug to the recipient for the first time, maximizing medication safety. The drug administration and dosage assessment system of this invention can obtain virtual pharmacokinetics data through virtual operation of the physiological pharmacokinetics virtual entity model, and combine this with the patient's actual pharmacokinetics data to determine the medication effect more accurately. Furthermore, in this invention, the virtual entity module and the twin data module can be iteratively updated according to changes in the data of the drug recipient, achieving co-growth and evolution with the drug recipient.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of medical informatics, specifically to a drug administration and dosage assessment system based on digital twins and its application. Background Technology

[0002] Physiological pharmacokinetic models can establish virtual populations of different age groups by predicting growth and organ maturation using mathematical models. By combining the absorption and disposition characteristics of drugs in healthy adult subjects with the established virtual populations, adult extrapolation can be performed to predict the pharmacokinetic characteristics of drugs in vivo. In recent years, physiological pharmacokinetic modeling has been used as an advanced method for determining optimal drug dosage in pediatric drug development.

[0003] However, the results of physiological pharmacokinetic models are the integrated results of a series of simulation processes. Each simulation is an abstraction of the complex processes in reality, and therefore, there may still be some errors compared to the actual situation. These errors will continue to accumulate and amplify, eventually leading to a large deviation between the predicted results and the measured values. On the other hand, there is still no effective solution to the uncertainty of the prediction results caused by the variability of relevant physiological parameters in specific populations.

[0004] Digital twin technology uses digital methods to create dynamic virtual models of physical entities across multiple dimensions, spatiotemporal scales, disciplines, and physical quantities. These models simulate and characterize the physical entity's attributes, behaviors, and rules in a real-world environment. A digital twin is a full-scale, full-lifecycle mapping of the physical entity; the two are identical and operate synchronously in both the physical and digital worlds. Based on the digital twin, the real-time state of the physical entity can be tracked throughout its lifecycle, and its state in specific environments can be simulated and predicted, enabling the digital twin to evolve alongside the physical entity.

[0005] Establishing a digital twin model of physiological pharmacokinetics and achieving complete mapping between virtual and biological entities, enabling real-time synchronous operation and evolution of virtual and biological entities under dynamic multidimensional and spatiotemporal scales, can greatly reduce the prediction error of the model. It can also create personalized prediction results for each individual to guide precision medicine. Summary of the Invention

[0006] In existing technologies, the use of physiological pharmacokinetic models to extrapolate to adults to predict the pharmacokinetic characteristics of drugs in vivo is prone to large errors and cannot effectively cope with the variability of physiological parameters.

[0007] This invention employs a method for establishing and applying a digital twin model of physiological pharmacokinetics, and proposes a method for determining the optimal drug dosage based on the digital twin model.

[0008] The technical solution of the present invention is as follows:

[0009] This invention provides a drug administration and dosage assessment system based on digital twins. The system includes the following modules: a basic information input module for drug administration subjects, a physiological pharmacokinetic physical entity data acquisition module, a physiological pharmacokinetic virtual entity module, a twin data module, and an output module.

[0010] The twin data module is connected to the drug administration subject basic information input module, the physiological pharmacokinetics physical entity data acquisition module, the physiological pharmacokinetics virtual entity module, and the output module, respectively.

[0011] The basic information entry module for drug administration subjects is used to collect basic information about the drug administration subjects and transmit this information to the twin data module. This basic information includes the ID information identifying the drug administration subject, as well as the subject's physiological information, including age, gender, and underlying disease information.

[0012] The physiological pharmacokinetic physical entity data acquisition module is used to collect pharmacokinetic data of the drug recipient and transmit the data to the twin data module.

[0013] The physiological pharmacokinetic virtual entity module is a digital simulation of the drug administration subject. It generates pharmacokinetic data of the drug administration subject through model simulation and transmits the generated data to the twin data module.

[0014] The twin data module is used to fuse the received data, determine whether to discontinue medication or provide a guide dose for the next medication, and output whether to discontinue medication or the guide dose for the next medication to the output module.

[0015] When administering the medication for the first time, the dosage assessment system outputs the dosage guidance using the following method:

[0016] (1) The twin data module generates the first dose of the drug subject based on the basic information of the drug subject, and transmits the first dose data to the physiological pharmacokinetics virtual entity module.

[0017] (2) The virtual entity module selects or creates a virtual entity model that is appropriate to the basic information of the drug administration object and performs virtual calculations.

[0018] (3) The twin data module evaluates the effect of the drug dosage simulation operation based on the data generated by the virtual operation of the physiological pharmacokinetics virtual entity module, and determines whether the dosage needs to be adjusted. If the dosage needs to be adjusted, the twin data module adjusts the dosage and sends the dosage back to the physiological pharmacokinetics virtual entity module to repeat steps (2)-(3) until the dosage does not need to be adjusted, and then pushes the first drug dosage to the output module.

[0019] When the medication is not being used for the first time, the dosage assessment system outputs the medication guidance dosage through the following method:

[0020] a. After the drug is administered to the recipient, the physiological pharmacokinetic physical entity data acquisition module collects the pharmacokinetic data of the recipient and transmits the collected data to the twin data module.

[0021] b. The physiological pharmacokinetic virtual entity performs virtual operation based on the drug dosage of the drug recipient, and transmits the obtained data to the twin data module;

[0022] c. The twin data module corrects the pharmacokinetic data of the obtained physiological pharmacokinetic virtual entity based on the pharmacokinetic data of its target drug, and further corrects the physiological pharmacokinetic virtual entity module.

[0023] d. The twin data module determines the effect of the previous drug dose on the drug recipient based on the pharmacokinetic data of the physical entity and the pharmacokinetic data of the virtual entity obtained from the operation. If the effect achieves the expected drug efficacy, or if the relevant physiological and biochemical indicators reach the threshold that must be stopped, the drug discontinuation instruction will be output to the output module. If the drug needs to be continued, the guidance dose for the next drug administration will be output to the output module.

[0024] In this invention, the twin data module includes: physical entity data of physiological pharmacokinetics, virtual entity data of physiological pharmacokinetics, service data, knowledge data, and fusion-derived data sub-modules. Virtual entity pharmacokinetics data obtained by the virtual entity module through virtual operation is stored in the physiological pharmacokinetics virtual entity data sub-module. Physical entity pharmacokinetics data collected by the physical entity data acquisition module is stored in the physiological pharmacokinetics physical entity data sub-module. The service data sub-module includes historical data, statistical data, and medical records, particularly adult physiological pharmacokinetics physical entity data and virtual entity data. The knowledge data sub-module includes expert knowledge, medical industry standards, treatment guidelines, and commonly used algorithm libraries. The fusion-derived data sub-module is derived data obtained by processing the aforementioned physical entity data of physiological pharmacokinetics, virtual entity data of physiological pharmacokinetics, service data, and knowledge data.

[0025] More specifically, in this invention, the physiological pharmacokinetic physical entity data submodule within the twin data module includes data that stores information reflecting the entity's function, state, and the operational status of its internal tissue, organ, and system levels. This data is acquired using sensing and detection technologies, originating from sensor data of the entity, including data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, as well as biochemical data such as blood routine tests, urine tests, and biological enzyme tests, especially the actual concentration of the drug in the blood. The physiological pharmacokinetic virtual entity data submodule includes various data generated from the virtual operation of the body model and drug model, as well as data generated from the interconnected operation of the drug model and body model, particularly the concentration data of the drug at the target site and in vital organs such as the heart, brain, liver, and kidneys, which are difficult to truly sense in the physical entity. The data submodule includes historical data, statistical data, and medical records, especially physical and virtual entity data related to adult physiological pharmacokinetics. The knowledge data submodule includes expert knowledge, medical industry standards, treatment guidelines, and commonly used algorithm libraries. The fusion and derivative data submodule is derived data obtained after data transformation, preprocessing, classification, association, integration, and fusion of the above-mentioned physical and virtual entity data, service data, and knowledge data. It uses technologies such as dynamic Bayesian networks and dynamic data-driven analysis and decision-making (DDDAS) to fuse physical reality data with multi-temporal and spatial related data, historical statistical data, expert knowledge, and other information data to obtain cyber-physical fusion data, thereby reflecting more comprehensive and accurate information and realizing information sharing and value-added services.

[0026] The fusion processing of the digital twin sub-modules described in this invention is achieved through the interaction of the aforementioned sub-modules. More specifically, dynamic Bayesian networks and dynamic data-driven analysis and decision-making techniques are widely used in digital twins. These techniques combine simulation and experiment / testing, allowing the simulation to dynamically receive data from the actual system and respond during execution. The simulation results can also dynamically control the operation of the actual system and guide the measurement process. The digital twin sub-modules described in this invention can be modeled based on existing neural network technology. The hidden layers of this neural network are generated through multiple iterations of calculations using service data and knowledge data. The aforementioned characteristic data, such as physiological changes at the target site of drug administration, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters, are used as the input layer data (X1, X2...X ... n After processing by a neural network algorithm, the output layer can output pharmacokinetic data, pharmacodynamic data, and guidance doses for the next administration (O1, O2...O...). n The fusion-derived data will be used to determine whether a new virtual run is needed, based on the algorithm-generated guideline dosage for the next medication.

[0027] In this invention, the physical entity data acquisition module for physiological pharmacokinetics is the physical entity pharmacokinetic data obtained by the drug administration subject through medical testing equipment after drug administration. The physical entity pharmacokinetic data includes blood drug concentration, CT, MRI, electrocardiogram, color Doppler ultrasound, etc., blood routine, urine test, biological enzyme data, and transmits the physical entity pharmacokinetic data to the twin data module.

[0028] The physiological pharmacokinetic physical entity data acquisition module comprises three levels of pharmacokinetic data acquisition sub-modules: unit level, subsystem level, and system level. The unit level, also known as the tissue level, includes medical data acquisition modules for at least one tissue, such as epithelial tissue, connective tissue, muscle tissue, and nervous tissue. The subsystem level, also known as the organ level, includes medical data acquisition modules for at least one organ, such as the heart, brain, liver, lungs, and kidneys. The system level includes medical data acquisition modules for at least one system, such as the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, and reproductive system.

[0029] In this invention, the physiological pharmacokinetic virtual entity module comprises two main models: an organism model and a drug model. The organism model includes four sub-models: an anatomical model, a tissue model, a physiological model, and a biochemical model. The drug model includes four sub-models: a physical model, a chemical model, a toxicological model, and a pharmacological model. The physiological pharmacokinetic virtual entity module can be constructed using software such as ANSYS, MATLAB, and WinNonlin, employing methods such as 3D scanning modeling, finite element analysis, edge computing, generative modeling, or model pruning. Furthermore, the physiological pharmacokinetic virtual entity must utilize neural network-based multiphysics model fitting techniques or particle swarm optimization algorithms to continuously iterate and grow through processes such as model fusion, model validation, model updating, and model management.

[0030] This invention further discloses a method for assessing drug dosage based on a digital twin-based drug dosage assessment system, comprising the following steps:

[0031] (1) The twin data module generates the first dose based on the basic information of the drug recipient and pushes the first dose to the physiological pharmacokinetics virtual entity module.

[0032] (2) The physiological pharmacokinetics virtual entity module generates running results based on the first dose or dose pushed by the twin data module and pushes the data to the twin data module.

[0033] (3) The twin data module evaluates the effect of the drug dosage simulation operation based on the data generated by the virtual operation results; determines whether the dosage needs to be adjusted. If the dosage needs to be adjusted, the twin data module adjusts the dosage and repeats steps (2)-(3); if the dosage does not need to be adjusted, the drug dosage is pushed to the output module to administer the drug to the drug recipient, and the drug dosage is transmitted to the physiological pharmacokinetic virtual entity module at the same time.

[0034] (4) After the drug is administered, the physiological pharmacokinetic physical entity data acquisition module collects the pharmacokinetic data of the drug recipient, obtains the physical entity pharmacokinetic data, and transmits it to the physical entity pharmacokinetic data in the twin data module; at the same time, the physiological pharmacokinetic virtual entity performs virtual operation according to the drug dosage pushed in step (3), and transmits the obtained data to the virtual entity pharmacokinetic data in the twin data module.

[0035] (5) The twin data module corrects the pharmacokinetic data of the physiological pharmacokinetic virtual entity based on its physical entity pharmacokinetic data, and further corrects the physiological pharmacokinetic virtual entity module.

[0036] (6) The twin data module judges the effect of the previous drug dose on the drug recipient based on the pharmacokinetic data of the physical entity and the pharmacokinetic data of the virtual entity. If the effect reaches the expected drug efficacy, or the relevant physiological and biochemical indicators reach the limit that must be stopped, the drug is stopped and the drug stop instruction is output to the output module. If it is necessary to continue the drug or adjust the dose, the guidance dose for the next drug is output to the output module, and then steps (2)-(6) are repeated.

[0037] In step (1), the initial medication dose is obtained based on the service data submodule in the twin data module.

[0038] In step (2), virtual entity physiological pharmacokinetic data are obtained through the interaction between the organism model and the drug model. The virtual entity physiological pharmacokinetic data includes drug concentration data at the target site, drug efficacy data, and toxicity in key areas such as the heart and brain.

[0039] In step (3), the twin data module determines the effect of the first dose simulation run by combining its service data and knowledge data sub-modules, and then determines whether the dose needs to be adjusted and the simulation run needs to be repeated by using the generated fusion-derived data.

[0040] In step (5), the twin data module performs data fusion processing by combining its service data and knowledge data sub-modules, and corrects the physiological pharmacokinetics virtual entity module based on the data processing results.

[0041] In step (6), the twin data module determines the effect of the medication based on the results of the data fusion processing described in step (5).

[0042] In a specific embodiment of the present invention, the drug administration dosage assessment and output are achieved through the following steps.

[0043] The first step is to use the physical and virtual entity data of adult physiological pharmacokinetics in the service data submodule of the twin data module as a basis, and based on the basic information of the drug administration subject, to extrapolate the initial dose of a specific drug by fusing derived data and using existing physiological pharmacokinetics algorithms and big data analysis technology, and then provide the initial dose to the physiological pharmacokinetics virtual entity module.

[0044] The second step involves inputting the initial dose into the physiological pharmacokinetics virtual entity module. The virtual entity module then performs a virtual operation to obtain virtual entity pharmacokinetics data. Through the interaction between the organism model and the drug model, the operation results are generated, and various key observation data are output to the physiological pharmacokinetics virtual entity data submodule in the twin data. In particular, data such as drug concentration data at the target site, drug efficacy data, and toxicity in key areas such as the heart and brain, which cannot be collected in the physiological pharmacokinetics physical entity data acquisition module, are included.

[0045] The third step involves obtaining data from the physiological pharmacokinetics virtual entity module of the twin data module. After obtaining the data from the physiological pharmacokinetics virtual entity module, the system combines the service data and knowledge data within it to determine the effect of the initial dose simulation. Then, through the generated fusion-derived data, it determines whether the dose needs to be adjusted and the simulation repeated, or whether it can be provided to the output module for administering the first dose to the drug recipient.

[0046] Fourth, if the fusion-derived data results indicate that the initial dose needs adjustment, the dose is readjusted, and steps two and three are repeated. If the fusion-derived data results indicate that the initial dose does not need adjustment, the data is provided to the output module to administer the initial dose to the recipient, and the initial dose is simultaneously input into the physiological pharmacokinetic virtual entity module.

[0047] The fifth step involves the body responding to the medication, with the response primarily represented by blood drug concentration. The pharmacokinetic physical entity data acquisition module collects pharmacokinetic data from the drug recipient. Through data sensing, it obtains data from necessary medical tests and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, as well as biochemical data such as complete blood counts, urinalysis, and enzyme tests. This data is then incorporated into the physiological pharmacokinetic physical entity data. Simultaneously, the physiological pharmacokinetic virtual entity module also runs virtually, generating corresponding results.

[0048] The sixth step involves obtaining the pharmacokinetic data of the drug recipients collected in the fifth step, as well as the pharmacokinetic data of the physiological pharmacokinetic virtual entity. This data is then combined with the service data and knowledge data to perform data fusion processing. Based on the data processing results, the physiological pharmacokinetic virtual entity module is corrected so that it can achieve the effect of synchronous evolution with the actual drug metabolism of the drug recipients.

[0049] Step 7: First, based on the data fusion results from the previous step, determine the actual effect of the previous medication dosage on the recipient. If the fusion data indicates that the recipient achieved the expected therapeutic effect, or if the relevant physiological or biochemical indicators reached the threshold requiring discontinuation of medication, then discontinue medication. If continued medication is necessary, output the guidance dosage for the next administration, and repeat steps 2 through 6.

[0050] In this invention, the physical entity refers to the drug administration target. In specific embodiments of this invention, children, patients with hepatic or renal insufficiency, the elderly, or pregnant women are the drug administration targets. In this invention, the physiological pharmacokinetic physical entity data acquisition module acquires data on blood drug concentration, CT scans, MRI, electrocardiograms, color Doppler ultrasound, complete blood counts, urinalysis, and bioenzyme data for children, patients with hepatic or renal insufficiency, the elderly, or pregnant women using sensors such as blood analyzers, X-ray machines, and CT scanners. In this invention, physiological pharmacokinetics refers to the drug metabolism kinetic data of the drug administration target after administration, or the drug metabolism kinetic data obtained by virtual entities based on model calculations.

[0051] The present invention further discloses a computer-readable storage medium storing a data processing program, which, when executed by a processor, implements the method for determining the optimal drug dosage based on a digital twin model.

[0052] The present invention, by employing the method for establishing and applying the digital twin model of physiological pharmacokinetics as described herein, can bring the following positive effects:

[0053] (1) After extrapolating the initial dose from the pharmacokinetic physical entity data and virtual entity data, the drug can be simulated and run on the physiological pharmacokinetic virtual entity module first. After the safety is proven, the drug can be administered to the subject for the first time to ensure the safety of the drug to the greatest extent.

[0054] (2) The effect of medication can be judged not only by the data actually sensed by the drug recipient, but also by the drug concentration data of the target site and important organs and the body model data of the target site and important organs in the virtual entity model of physiological pharmacokinetics, so that the judgment results and timing are more accurate.

[0055] (3) The dosage of each medication is determined sequentially based on the drug metabolism kinetic data collected from the drug recipient in the previous administration and the results of the operation of the physiological pharmacokinetics virtual entity module, so as to achieve precise medication.

[0056] (4) For each specific individual, a personalized physiological pharmacokinetics virtual entity and twin data model can be established. This personalized physiological pharmacokinetics virtual entity and twin data model can grow and evolve together with the specific individual, so that the results of the virtual operation can more realistically reflect the state of the drug user. Attached Figure Description

[0057] Figure 1 This is a diagram showing the overall structure of the digital twin model of physiological pharmacokinetics.

[0058] Figure 2 Structural diagram of the physical entity information acquisition module for pediatric physiological pharmacokinetics;

[0059] Figure 3 A diagram of the virtual entity module structure for pediatric physiological pharmacokinetics;

[0060] Figure 4 A structural diagram of the twin data module in a digital twin model of pediatric physiological pharmacokinetics;

[0061] Figure 5 A schematic diagram of a digital twin model application system for children's physiological pharmacokinetics;

[0062] Figure 6 Flowchart for the application of digital twin models of pediatric physiological pharmacokinetics;

[0063] Figure 7 This is a block diagram illustrating the principle of neural network analysis technology.

[0064] Figure 8 A structural diagram of the physical entity information acquisition module for physiological pharmacokinetics in patients with hepatic and renal insufficiency;

[0065] Figure 9 A virtual entity module structure diagram for the physiological pharmacokinetics of patients with hepatic and renal insufficiency;

[0066] Figure 10 A diagram showing the structure of the twin data module in the digital twin model of physiological pharmacokinetics in patients with hepatic and renal insufficiency;

[0067] Figure 11 A schematic diagram of the application system of a digital twin model of physiological pharmacokinetic dynamics for patients with liver and kidney dysfunction;

[0068] Figure 12 Structural diagram of the physical entity information acquisition module for physiological pharmacokinetics in the elderly;

[0069] Figure 13 A structural diagram of a virtual entity module for physiological pharmacokinetics in the elderly;

[0070] Figure 14 A structural diagram of the twin data module in a digital twin model of physiological pharmacokinetics in the elderly;

[0071] Figure 15 A schematic diagram of the application system of digital twin model of physiological pharmacokinetics in the elderly;

[0072] Figure 16 A structural diagram of the physical entity information acquisition module for pregnant women's physiological pharmacokinetics.

[0073] Figure 17 A diagram of the virtual entity module structure for the physiological pharmacokinetics of pregnant women;

[0074] Figure 18 A structural diagram of the twin data module in the digital twin model of physiological pharmacokinetics in pregnant women;

[0075] Figure 19 Flowchart for establishing a virtual entity model of physiological pharmacokinetics in pregnant women;

[0076] Figure 20 A schematic diagram of the application system of digital twin model for the physiological pharmacokinetics of pregnant women. Detailed Implementation

[0077] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0078] Example 1: Digital Twin Dosage Assessment System for Children

[0079] Children are in a period of growth and development, and their bodily functions are not yet fully developed. Their immunity is relatively low, and their liver and kidneys' detoxification and excretion functions, as well as the blood-brain barrier, are not fully mature. Improper medication can have a significant impact on these developing organs. Currently, the lack of or incomplete information on pediatric medication use in drug instructions remains a prominent issue. Dosage for children is often determined based on physician experience, leading to frequent off-label use. Determining safe and effective dosages remains a challenge in pediatric drug development trials. This embodiment employs a digital twin model of pediatric physiological pharmacokinetics and presents a method for determining the optimal pediatric medication dosage based on this model.

[0080] The present invention describes a method for establishing and applying a digital twin model of pediatric physiological pharmacokinetics. The first step is to establish a digital twin model of pediatric physiological pharmacokinetics, which comprises five main parts: a pediatric basic information input module, a pediatric physiological pharmacokinetics physical entity data acquisition module, a pediatric physiological pharmacokinetics virtual entity module, a pediatric twin data module, and an output module. The pediatric physiological pharmacokinetics physical entity data acquisition module obtains comprehensive information about the child as the drug administration subject through medical detection methods via sensing devices, maps this information to the pediatric physiological pharmacokinetics virtual entity module, and updates the twin data module. The pediatric physiological pharmacokinetics virtual entity module and the twin data module operate virtually and iteratively optimize each other under interactive conditions.

[0081] The pediatric physiological pharmacokinetic physical entity data acquisition module collects physical pharmacokinetic data from medical testing equipment after drug administration. This physical pharmacokinetic data includes blood drug concentration, CT scans, MRI, electrocardiograms, color Doppler ultrasound, complete blood counts, urinalysis, and bioenzyme data. This physical pharmacokinetic data is then transmitted to the twin data module. This module collects drug metabolism data at three levels: unit level, subsystem level, and system level. The unit level, also known as the tissue level, includes epithelial tissue, connective tissue, muscle tissue, and nervous tissue. Above the unit level is the subsystem level, also known as the organ level, including organs such as the heart, brain, liver, lungs, and kidneys. Above the subsystem level is the system level, including the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, and reproductive system. The above three levels of systems constitute an organism, which is organically integrated with each other. During its operation, data such as body temperature, blood pressure, blood oxygen, electrocardiogram, imaging data of some organs, laboratory data, height, and weight can be obtained through certain equipment. However, most data that characterize the health status of various tissues and the status of the system, such as plasma protein binding rate, gastrointestinal physiological parameters, liver clearance rate, glomerular filtration rate, and creatinine clearance rate, cannot be directly measured. These parameters are what pharmacokinetics is more concerned with. Therefore, it is necessary to rely on the establishment of a virtual entity model of pediatric physiological pharmacokinetics and directly obtain virtual parameters from the model.

[0082] The pediatric physiological pharmacokinetics virtual entity module comprises two main models: a pediatric organism model and a drug model. The pediatric organism model includes four parts: anatomical, tissue, physiological, and biochemical models. The drug model includes four parts: physical, chemical, toxicological, and pharmacological models. The pediatric physiological pharmacokinetics virtual entity module can be constructed using software such as ANSYS, MATLAB, and WinNonlin, employing methods such as 3D scanning modeling, finite element analysis, edge computing, generative modeling, and model pruning. Furthermore, the module must utilize neural network-based multiphysics model fitting techniques or particle swarm optimization algorithms, continuously iterating and evolving through model fusion, validation, updating, and management. The model updating process is particularly crucial. When discrepancies exist between the virtual model's results and the physical objects generated by the physical entity's operation, and the results fail to meet requirements, the model must be updated to more closely approximate the actual state or characteristics of the physical objects. It utilizes methods such as 3D scanning modeling, finite element analysis, edge computing, generative modeling, and model pruning to construct models; and performs model fusion, validation, updating, and management based on neural network multiphysics model fitting techniques or particle swarm optimization algorithms. Figure 7 Taking the neural network analysis technique shown in the figure as an example, in this embodiment, the hidden layer of the neural network is generated by establishing and iterating through service data and knowledge data. The aforementioned physiological changes of the target site in children, the protein binding rate of children, gastrointestinal physiological parameters, liver clearance rate, kidney function parameters, and other feature data are used as the input layer data (X1, X2...X...) of the neural network. n After processing by a neural network algorithm, the output layer can output pharmacokinetic data, pharmacodynamic data, and guidance doses for the next administration (O1, O2...O...). n The fused derived data will be used to determine whether a new virtual run is needed based on the next medication guidance dosage generated by the algorithm. The neural network algorithm used in this embodiment is a publicly available method.

[0083] The twin data module comprises a physical entity data submodule for pediatric physiological pharmacokinetics, a virtual entity data submodule for pediatric physiological pharmacokinetics, service data, knowledge data, and fused derived data. The physical entity data for pediatric physiological pharmacokinetics refers to data that reflects the function, state, and operational status of various parts of the entity at the tissue, organ, and system levels. This data originates from sensor data of the entity, including data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, as well as biochemical data such as blood routine tests, urine tests, and biological enzymes. In the twin data of this invention, the focus is on data related to the pharmacokinetic characteristics of drugs in the human body, such as data on absorption, distribution, metabolism, and excretion. Only a small portion of this data is actually measurable, such as height and weight, drug concentration in the child's blood, and blood flow velocity; the rest is difficult to measure.

[0084] The data in the pediatric physiological pharmacokinetics virtual entity data submodule does not depend on the drug administration object, but directly obtains data from the mathematical model within the pediatric physiological pharmacokinetics virtual entity module. This includes various data generated by the virtual operation of the pediatric body model and drug model, as well as data generated by the interrelation between the drug model and the pediatric body model. In particular, data directly related to pharmacokinetic characteristics, such as the shape and size of the target organ, the blood flow of the target organ, the concentration of the drug at the target site, the expression level of enzymes, the plasma protein binding rate, the volume of distribution, and the clearance rate, are data that are difficult to truly sense in a physical entity.

[0085] The service data submodule includes historical data, statistical data, and medical records, especially physical and virtual entity data related to adult physiological pharmacokinetics. The knowledge data submodule includes expert knowledge, medical industry standards, rule constraints, and commonly used algorithm libraries. The fusion-derived data submodule is derived data obtained after data transformation, preprocessing, classification, association, integration, and fusion of the aforementioned physical and virtual entity data, service data, and knowledge data related to pediatric physiological pharmacokinetics. It employs technologies such as dynamic Bayesian networks and dynamic data-driven analysis and decision-making (DDDAS) to obtain cyber-physical fusion data by fusing physical reality data with multi-temporal and spatial related data, historical statistical data, and expert knowledge. This results in more comprehensive and accurate information and enables information sharing and value-added services. The fusion-derived data includes guiding dosage data for pediatric medication, judgmental data on actual medication outcomes for pediatric patients, judgmental data on the virtual operation results of the pediatric physiological pharmacokinetics virtual entity module, data that self-updates the twin data itself, and data that updates the pediatric physiological pharmacokinetics virtual entity module.

[0086] The following is a brief description of how the digital twin model of pediatric physiological pharmacokinetics described in this invention can be used to determine the dosage of a certain drug in a child's body.

[0087] The first step, based on the adult physiological pharmacokinetic physical entity data and virtual entity data in the service data submodule of the twin data, involves fusing derived data and utilizing existing physiological pharmacokinetic algorithms and big data analytics to extrapolate the initial dose of specific pediatric medications. Figure 5 The information flow shown in section ② is to input the initial dose into the pediatric physiological pharmacokinetics virtual entity.

[0088] The second step involves a virtual entity module for pediatric physiological pharmacokinetics. Figure 5 After receiving the initial dose, the information flow ② shown in the diagram first performs virtual operation, through... Figure 5 The information flow shown in ③ ensures the interaction between the pediatric organism model and the drug model, generates operational results, and transmits various key observation data through... Figure 5 The information flow shown in ④ is output to the pediatric physiological pharmacokinetics virtual entity data submodule in the twin data, especially data that cannot be sensed in the drug administration subject, such as drug concentration data at the target site, plasma protein binding rate, liver clearance rate, glomerular filtration rate, and creatinine clearance rate.

[0089] The third step involves the virtual entity data submodule of the twin data for children's physiological pharmacokinetics. Figure 5 As shown in information flow ④, after obtaining the data from the virtual entity module for pediatric physiological pharmacokinetics, the focus is on characteristic data such as physiological changes at the target site in children, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters. This data is then combined with service and knowledge data from the twin database and utilized... Figure 5 Information flow ⑤ shown in the figure uses data mining and analysis techniques to extract useful information from the original data. Based on the use of expert knowledge, rule base information and other data, it generates fused derivative data. It uses neural network analysis technology, dynamic Bayesian network and dynamic data-driven analysis and decision-making technology (DDDAS technology) to output the effect of the first dose virtual run.

[0090] Fourth, if the results of the fusion and derived data indicate that the previous virtual run of the pediatric physiological pharmacokinetics virtual entity module requires dose adjustment and re-run, then... Figure 5 The information flow shown in step ② re-outputs the guidance dose to the pediatric physiological pharmacokinetics virtual entity module, and then repeats steps two and three above. If the fusion-derived data results determine that the previous virtual run of the pediatric physiological pharmacokinetics virtual entity did not require dose adjustment, then... Figure 5The information flow shown in section ① provides the confirmed dose to the output module, and administers the medication to the patient based on the confirmed dose, while simultaneously... Figure 5 The information flow shown in section ② inputs the confirmed dose into the pediatric physiological pharmacokinetics virtual entity submodule.

[0091] Fifth step, based on the output module... Figure 5 The information flow shown in section ① obtains the drug dosage and administers the drug. After the patient receives the drug, the body produces corresponding bodily responses, represented by physiological changes, changes in target site function, and changes in blood drug concentration. Data is then transmitted through sensing to obtain data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, or biochemical data such as complete blood counts, urine tests, and enzyme tests. All of this data is then processed through… Figure 5 The information flow shown is directed to the physical entity data of pediatric physiological pharmacokinetics provided by ⑥. Simultaneously, the virtual entity of pediatric physiological pharmacokinetics also runs virtually, generating corresponding results, and... Figure 5 The information flow shown in diagram ④ is provided to the virtual entity data submodule for children's physiological pharmacokinetics.

[0092] Step 6, twin data through Figure 5 The information flows shown in ⑥ and ④, after obtaining the pharmacokinetic data of the drug recipients and the operational data of the virtual entity module of pediatric physiological pharmacokinetics generated in step 5, respectively, focus on comparing the differences between the physical entity data and the virtual entity data of pediatric physiological pharmacokinetics. Then, combining this with the service data and knowledge data in the twin data, and utilizing... Figure 5 The information flow shown in section ⑤ undergoes data fusion processing, and then, combined with expert knowledge, rule base information, and other data, fusion-derived data is generated to produce result determination data for the previous medication dosage on the patient. On the other hand, based on the fusion-derived data, through... Figure 5 The information flow shown in section ⑦ corrects the virtual entity module of pediatric physiological pharmacokinetics, enabling the virtual entity module of pediatric physiological pharmacokinetics to achieve the effect of synchronous evolution with the drug metabolism of pediatric drug users.

[0093] Step 7: If the previous medication dosage, based on the drug metabolism data of the patient, indicates that the patient has achieved the expected therapeutic effect and can discontinue the medication, or if relevant physiological and biochemical indicators have reached the threshold for discontinuing the medication, then the medication is discontinued. If continued medication is required, the fusion-derived data will output the guiding dosage for the next medication, and steps 2 through 6 above will be repeated.

[0094] By adopting the aforementioned methods for establishing and applying digital twin models of pediatric physiological pharmacokinetics, the safety of pediatric medication can be ensured to the greatest extent possible, and precise control of dosage can be achieved for each administration. Furthermore, if a personalized virtual entity and twin data model of pediatric physiological pharmacokinetics can be established for each individual child, this personalized virtual entity and twin data model can grow and evolve alongside the specific child. This not only allows the results of virtual operation to more realistically reflect the individual's actual state but also provides more medical service support throughout the child's lifespan.

[0095] Physiological pharmacokinetic models in children can predict their growth and organ maturation through mathematical models, establishing virtual populations of different age groups. By combining the absorption and disposition characteristics of drugs in healthy adult subjects with the established virtual pediatric populations, extrapolation from adults to children can be achieved to predict the pharmacokinetic characteristics of drugs in children. In recent years, physiological pharmacokinetic modeling has been used in pediatric drug development as an advanced method for determining optimal pediatric drug dosages. However, current techniques using physiological pharmacokinetic models to extrapolate from adults to children to predict drug pharmacokinetic characteristics in children are prone to significant errors and cannot effectively address the variability of children's physiological parameters.

[0096] Example 2: Digital Twin Dosage Assessment System for Patients with Hepatic and Renal Impairment

[0097] The liver and kidneys are the primary sites of drug metabolism and excretion; therefore, hepatic and renal insufficiency affects the entire pharmacokinetic process of drugs. Impaired liver function has a significant impact on orally administered drugs with a first-pass effect, increasing blood drug concentration and bioavailability. It can also reduce the plasma protein binding rate of most drugs, increasing the concentration of free drug, thereby increasing efficacy or even causing toxic effects. Due to a significant reduction in the amount or activity of hepatic drug-metabolizing enzymes, the rate and extent of drug metabolism through these enzymes are significantly slowed, leading to increased parent drug concentration and prolonged elimination half-life, thus increasing efficacy or even causing toxic effects. For drugs primarily eliminated via renal excretion, renal damage can alter pharmacokinetics and efficacy. The most obvious change caused by renal damage is a decrease in the renal secretion or excretion of the drug or its metabolites. Renal damage can also cause changes in drug absorption, hepatic metabolism, plasma protein binding, and drug distribution. These changes may be particularly pronounced in patients with severe renal impairment, and can even be observed when the kidneys are not the primary route of drug excretion. Ensuring medication safety for patients with hepatic and renal insufficiency requires comprehensive consideration of the patient's liver and kidney function, disease status, dosage, type of medication, and administration method. Dosage is a crucial factor affecting efficacy and reducing drug side effects; controlling the dosage can achieve therapeutic effects while minimizing the likelihood of adverse reactions. Currently, dosage selection is mostly based on physician experience, and determining a safe and effective dosage remains a challenge. This embodiment employs a method for establishing and applying a digital twin model of the physiological pharmacokinetics of patients with hepatic and renal insufficiency, and proposes a method for determining the optimal medication dosage for these patients based on this digital twin model.

[0098] The digital twin model of physiological pharmacokinetics for patients with hepatic and renal insufficiency comprises five main parts: a basic information input module for patients with hepatic and renal insufficiency, a physical entity data acquisition module for physiological pharmacokinetics of patients with hepatic and renal insufficiency, a virtual entity model for physiological pharmacokinetics of patients with hepatic and renal insufficiency, a twin data module, and an output module. The physical entity data acquisition module acquires data through sensing devices, maps it to the virtual entity model for physiological pharmacokinetics of patients with hepatic and renal insufficiency, and updates the twin data. The virtual entity model and the twin data module operate virtually and iteratively optimize each other under interactive conditions. These three parts—the physical entity data acquisition module, the virtual entity model, and the twin data module—are interconnected, mutually optimizing and iteratively updating, and evolving synchronously.

[0099] The physical entity data acquisition module for physiological pharmacokinetics in patients with hepatic and renal insufficiency obtains physical entity pharmacokinetic data from medical testing equipment after drug administration. This physical entity pharmacokinetic data includes blood drug concentration, CT scans, MRI, electrocardiograms, color Doppler ultrasound, complete blood counts, urinalysis, and bioenzyme data. This physical entity pharmacokinetic data is transmitted to the twin data module. It comprises three levels: unit level, subsystem level, and system level. The unit level, also known as the tissue level, includes epithelial tissue, connective tissue, muscle tissue, and nervous tissue. Above the unit level is the subsystem level, also known as the organ level, including organs such as the liver, kidneys, heart, brain, gastrointestinal tract, gallbladder, and lungs. Above the subsystem level is the system level, including the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, and reproductive system. The above three levels of systems constitute an organism, which is organically integrated with each other. During its operation, data such as body temperature, blood pressure, blood oxygen, electrocardiogram, imaging data of some organs, laboratory data, height, and weight can be obtained through certain equipment. However, most data that characterize the health status of various tissues and the status of the system, such as plasma protein binding rate, gastrointestinal physiological parameters, liver clearance rate, glomerular filtration rate, and creatinine clearance rate, cannot be directly measured. These parameters are what pharmacokinetics are more concerned with. Therefore, it is necessary to rely on the establishment of a virtual entity of physiological pharmacokinetics for patients with liver and kidney dysfunction and to directly obtain virtual parameters from the model.

[0100] The virtual entity module for the physiological pharmacokinetics of patients with hepatic and renal insufficiency comprises two main models: a patient organism model and a drug model. The organism model includes six parts: anatomical, tissue, physiological, biochemical, liver, and kidney function models. The drug model includes four parts: physical, chemical, toxicological, and pharmacological models. The virtual entity module for the physiological pharmacokinetics of patients with hepatic and renal insufficiency undergoes continuous iteration and growth through processes such as model construction, model assembly, model fusion, model validation, model updating, and model management. The model updating process is particularly crucial. It addresses situations where discrepancies exist between the virtual model's results and the physical objects generated by the physical entity's operation, or when the results fail to meet requirements. In such cases, the model must be updated to more closely approximate the actual state or characteristics of the physical objects.

[0101] The twin data module includes physical entity data on the physiological pharmacokinetics of patients with hepatic and renal insufficiency, virtual entity data on the physiological pharmacokinetics of patients with hepatic and renal insufficiency, service data, knowledge data, and fused derived data. The physical entity data on the physiological pharmacokinetics of patients with hepatic and renal insufficiency refers to data that reflects the physical function and state of the patient, as well as the working conditions of its internal tissues, organs, and systems. This data comes from sensor data of the entity, including data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, and biochemical data such as blood routine tests, urine tests, and biological enzymes. In the twin data of this invention, the focus is on data related to the pharmacokinetic characteristics of drugs in the human body, such as data on absorption, distribution, metabolism, and excretion. Only a small portion of this data is actually measurable, such as height and weight, drug concentration in the blood of patients with hepatic and renal insufficiency, and blood flow velocity; other data are difficult to measure.

[0102] The virtual entity data for the physiological pharmacokinetics of patients with hepatic and renal insufficiency does not rely on the data obtained from the physical entity data acquisition module for the physiological pharmacokinetics of patients with hepatic and renal insufficiency. Instead, it directly obtains data from the mathematical model within the virtual entity module. This includes various data generated by the virtual operation of the patient's body model and drug model, as well as data generated by the interaction between the drug model and the patient's body model. In particular, data directly related to pharmacokinetic characteristics, such as the shape and size of the target organ, the blood flow to the target organ, the concentration of the drug at the target site, the expression level of enzymes, the plasma protein binding rate, the volume of distribution, and the clearance rate, are difficult to truly sense in a physical entity.

[0103] Service data includes historical data, statistical data, and medical records, especially data collected by the physical entity data acquisition module for adult physiological pharmacokinetics and virtual entity data. Knowledge data includes expert knowledge, treatment guidelines, rule constraints, and commonly used algorithm libraries. Fusion-derived data is derived from the above-mentioned physical entity data, virtual entity data, service data, and knowledge data for patients with hepatic and renal insufficiency after data transformation, preprocessing, classification, association, integration, and fusion. By fusing physical reality data with multi-temporal and spatial related data, historical statistical data, expert knowledge, and other information data, information-physical fusion data is obtained, thereby reflecting more comprehensive and accurate information and realizing information sharing and value-added. Fusion-derived data includes guiding dosage data for medication in patients with hepatic and renal insufficiency, judgmental data on actual medication results obtained by the physical entity data acquisition module for patients with hepatic and renal insufficiency, judgmental data on the virtual operation results of the virtual entity module for patients with hepatic and renal insufficiency, data that self-updates the twin data itself, and data that updates the virtual entity module for patients with hepatic and renal insufficiency.

[0104] The following is combined Figure 11 The diagram shown is a "Schematic diagram of the application system of the digital twin model of physiological pharmacokinetic dynamics in patients with hepatic and renal insufficiency". Figure 6 The "Flowchart for Application of Physiological Pharmacokinetic Digital Twin Model" shown describes how to use the physiological pharmacokinetic digital twin model for patients with hepatic and renal insufficiency as described in this invention to determine the dosage of a certain drug in the body of patients with hepatic and renal insufficiency.

[0105] The first step, based on the adult physiological pharmacokinetic physical entity data and virtual entity data in the "Service Data" section of the twin data module, extrapolates the initial dose of medication for specific patients with hepatic or renal insufficiency by fusing derived data, and uses... Figure 11 The information flow shown in section ② inputs the initial dose to the "physiological pharmacokinetics virtual entity of patients with hepatic and renal insufficiency".

[0106] The second step involves using a virtual entity representing the physiological pharmacokinetics of patients with hepatic and renal insufficiency. Figure 11 After receiving the initial dose, the information flow ② shown in the diagram first performs virtual operation, through... Figure 11 The information flow shown in ③ ensures the interaction between the body model and the drug model of patients with liver and kidney dysfunction, generates operational results, and transmits various key observation data through... Figure 11The information flow shown in ④ is output to the "physiological pharmacokinetic virtual entity data of patients with hepatic and renal insufficiency" in the twin data, especially the drug concentration data at the target site, plasma protein binding rate, liver clearance rate, glomerular filtration rate, creatinine clearance rate and other data that cannot be sensed in the physical entity data acquisition module of the physiological pharmacokinetic data of patients with hepatic and renal insufficiency.

[0107] The third step involves using virtual entity data on the physiological pharmacokinetics of patients with hepatic and renal insufficiency from twin data. Figure 11 The information flow shown in ④, after obtaining the data from the virtual entity operating on the physiological pharmacokinetics of patients with hepatic and renal insufficiency, focuses on characteristic data such as target site physiological changes, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters. This data is then combined with service and knowledge data from the twin database and utilized... Figure 11 The information flow shown in section ⑤ employs data mining and analysis techniques to extract useful information from the original data. Based on expert knowledge, rule base information, and other data, it generates fused derived data, outputting the effect of the initial dose virtual run. The effect of the virtual run determines whether dose adjustment and re-run are necessary. If adjustment and re-run are required, the fused derived data also outputs a guiding dose for the next virtual run. Specifically, neural network analysis, dynamic Bayesian networks, and Dynamic Data Driven Analysis and Decision Support (DDDAS) techniques can be used to fuse the derived data. More specifically, in this embodiment, the hidden layer of the neural network is built from service data and knowledge data and generated through multiple iterative calculations. The aforementioned characteristic data such as physiological changes at the target site, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters of patients with liver and kidney dysfunction are used as the input layer data (X1, X2...X ... n After processing by a neural network algorithm, the output layer can output pharmacokinetic data, pharmacodynamic data, and guidance doses for the next administration (O1, O2...O...). n The fusion-derived data will be used to determine whether a new virtual run is needed, based on the algorithm-generated guideline dosage for the next medication.

[0108] Fourth, if the fusion of derived data determines that the virtual run of the physiological pharmacokinetics virtual entity for patients with hepatic and renal insufficiency needs to be re-run with dosage adjustments, then... Figure 11 The information flow shown in step ② re-outputs the guidance dose to the "virtual entity of physiological pharmacokinetics for patients with hepatic and renal insufficiency," and then repeats steps two and three above. If the fusion-derived data results determine that the previous virtual run of the "virtual entity of physiological pharmacokinetics for patients with hepatic and renal insufficiency" does not require dose adjustment, then... Figure 11The information flow shown in section ① is to provide the confirmed dose to patients with hepatic or renal insufficiency, while simultaneously... Figure 11 The information flow shown in section ② inputs the confirmed dose into the physiological pharmacokinetic virtual entity of the patient with hepatic and renal insufficiency.

[0109] Step 5: Physiological pharmacokinetics physical data acquisition module for patients with hepatic and renal insufficiency. Figure 11 The information flow shown ① indicates the drug dosage is obtained and administered. After receiving the drug, patients with liver and kidney dysfunction experience corresponding bodily responses, represented by physiological changes, changes in target site function, and changes in blood drug concentration. Data is then transmitted, obtaining data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, or biochemical data such as complete blood counts, urinalysis, and enzyme tests. This data is then processed through… Figure 11 The information flow shown is directed to the physical entity data of the physiological pharmacokinetics of patients with hepatic and renal insufficiency provided by ⑥. Simultaneously, the virtual entity of the physiological pharmacokinetics of patients with hepatic and renal insufficiency also runs virtually, generating corresponding results, and... Figure 11 The information flow shown in ④ is provided to the virtual entity data of physiological pharmacokinetics of patients with liver and kidney dysfunction.

[0110] Step 6, twin data through Figure 11 The information flows shown in ⑥ and ④ respectively obtain the data collected by the physical entity data acquisition module for the physiological pharmacokinetics of patients with liver and kidney dysfunction generated in step 5, and the data obtained from the operation of the virtual entity for the physiological pharmacokinetics of patients with liver and kidney dysfunction. The key focus is on comparing the differences between the physical entity data and the virtual entity data for the physiological pharmacokinetics of patients with liver and kidney dysfunction. This is then combined with the service data and knowledge data in the twin data, utilizing... Figure 11 The information flow shown in section ⑤ undergoes data fusion processing, and then, combined with expert knowledge, rule base information, and other data, fusion-derived data is generated to produce result determination data for the previous medication dosage in patients with hepatic and renal insufficiency. On the other hand, based on the fusion-derived data, through... Figure 11 The information flow shown in section ⑦ corrects the "physiological pharmacokinetics virtual entity of patients with liver and kidney dysfunction" so that the "physiological pharmacokinetics virtual entity of patients with liver and kidney dysfunction" can evolve synchronously with the patients with liver and kidney dysfunction.

[0111] Step 7: If the data collected by the physiological pharmacokinetic physical entity data acquisition module for patients with hepatic and renal insufficiency indicates that the patient has achieved the expected therapeutic effect of the drug and can discontinue the medication, or if the relevant physiological and biochemical indicators have reached the threshold for discontinuing the medication, then the medication is discontinued. If continued medication is required, the fused derivative data will output the guiding dose for the next administration, and steps 2 through 6 above will be repeated.

[0112] By adopting the aforementioned methods for establishing and applying the physiological pharmacokinetic digital twin model for patients with hepatic and renal insufficiency, the safety of medication use in these patients can be maximized, and precise control of dosage can be achieved for each administration. Furthermore, if a personalized "physiological pharmacokinetic virtual entity" and "twin data" can be established for each individual patient with hepatic and renal insufficiency, these personalized entities can develop and evolve alongside that individual. This not only allows the virtual operation results to more realistically reflect the patient's condition but also provides greater support for healthcare services throughout their lifespan.

[0113] Example 3: Digital Twin Dosage Assessment System for the Elderly

[0114] As people age, their organs undergo changes, and physiological functions gradually decline, leading to alterations in the absorption, distribution, metabolism, and excretion of drugs. The number of gastrointestinal mucosal cells decreases in older adults, digestive motility slows, gastric acid secretion decreases, and intestinal arteriosclerosis reduces blood flow, thus affecting drug absorption and preventing optimal therapeutic effects. Reduced liver blood flow and slower kidney excretion significantly impair liver detoxification and drug metabolism, causing drugs to accumulate in the body, increasing drug sensitivity and the risk of toxic reactions. The decrease in body water and muscle tissue, coupled with an increase in fat content, alters drug distribution and reduces tolerance. Individual differences among older adults vary considerably across age groups, resulting in vastly different dosages even within the same age group. Many older adults suffer from multiple chronic diseases and often require the simultaneous and long-term use of various medications, increasing the risk of drug interactions, adverse events, and even serious adverse reactions. This makes medication management in the elderly even more challenging. As a special group, the elderly have unique characteristics in drug absorption, distribution, metabolism, and excretion within the body, thus requiring lower dosages than the average adult. However, while drug instructions often specify adult and pediatric dosages, they often fail to indicate dosages for the elderly. Currently, drug dosages for the elderly are mostly determined based on the experience of physicians or pharmacists, and determining a safe and effective dosage remains a challenge.

[0115] This embodiment employs a method for establishing and applying a digital twin model of physiological pharmacokinetics in the elderly, and proposes a method for determining the optimal medication dosage for the elderly based on the digital twin model.

[0116] The overall structure of the digital twin model of physiological pharmacokinetics in the elderly comprises five main parts: a module for inputting basic information on patients with hepatic and renal insufficiency; a physical entity data acquisition module for physiological pharmacokinetics in the elderly; a virtual entity for physiological pharmacokinetics in the elderly; a twin data module; and an output module. The physical entity data acquisition module acquires data through sensing devices, maps it to the virtual entity module, and updates the twin data. The virtual entity module and the twin data module interact and operate virtually, mutually optimizing and iterating. These three parts—the physical entity data acquisition module, the virtual entity module, and the twin data module—are interconnected, mutually optimizing and iterating, and evolving synchronously.

[0117] The physical entity data acquisition module for physiological pharmacokinetics in the elderly obtains physical pharmacokinetic data from medical testing equipment after drug administration. This physical pharmacokinetic data includes blood drug concentration, CT scans, MRI, electrocardiograms, color Doppler ultrasound, complete blood counts, urinalysis, and bioenzyme data. This physical pharmacokinetic data is then transmitted to the twin data module. It comprises three levels: unit level, subsystem level, and system level. The unit level, also known as the tissue level, includes epithelial tissue, connective tissue, muscle tissue, and nervous tissue. Above the unit level is the subsystem level, also known as the organ level, which includes organs such as the heart, brain, liver, gastrointestinal tract, spleen, lungs, kidneys, muscles, bones, and blood vessels. Above the subsystem level is the system level, which includes the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, and reproductive system. The above three levels of systems constitute an organism, which is organically integrated with each other. During its operation, data such as body temperature, blood pressure, blood oxygen, electrocardiogram, imaging data of some organs, laboratory data, height, and weight can be obtained through certain equipment. However, most data that characterize the health status of various tissues and the status of the system, such as plasma protein binding rate, gastrointestinal physiological parameters, liver clearance rate, glomerular filtration rate, and creatinine clearance rate, cannot be directly measured. These parameters are what pharmacokinetics is more concerned with. Therefore, it is necessary to rely on the establishment of a virtual entity of physiological pharmacokinetics in the elderly and directly obtain virtual parameters from the model.

[0118] The virtual entity module for the physiological pharmacokinetics of the elderly comprises two main models: an elderly organism model and a drug model. The organism model includes five parts: anatomical model, tissue model, physiological model, biochemical model, and underlying disease model. The drug model includes four parts: physical model, chemical model, toxicological model, and pharmacological model. The virtual entity module for the physiological pharmacokinetics of the elderly must continuously iterate and grow through processes such as model construction, model assembly, model fusion, model validation, model updating, and model management. The model updating process is particularly critical. It refers to updating the model when discrepancies exist between the results of the virtual model operation and the physical objects generated by the physical entity operation, or when the model fails to meet requirements, to make the model more closely approximate the actual state or characteristics of the physical objects.

[0119] The twin data system comprises physical entity data on the physiological pharmacokinetics of the elderly, virtual entity data on the physiological pharmacokinetics of the elderly, service data, knowledge data, and fused derived data. The physical entity data on the physiological pharmacokinetics of the elderly refers to data that reflects the physical function and state of the elderly, as well as the working status of various parts of the entity at the tissue, organ, and system levels. This data comes from sensor data of the entity, including data from medical tests and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, and biochemical data such as blood routine tests, urine tests, and biological enzymes. In the twin data of this invention, the focus is on data related to the pharmacokinetic characteristics of drugs in the human body, such as data on absorption, distribution, metabolism, excretion, and drug interactions. Only a small portion of this data is actually measurable, such as height and weight, drug concentration in the blood of the elderly, and blood flow velocity; other data are difficult to measure.

[0120] The virtual entity data on physiological pharmacokinetics of the elderly does not rely on the data obtained from the physical entity data acquisition module for physiological pharmacokinetics of the elderly, but directly obtains data from the mathematical model within the virtual entity. This includes various data generated by the virtual operation of the elderly body model and drug model, as well as data generated by the interaction between the drug model and the elderly body model. In particular, data directly related to pharmacokinetic characteristics, such as the shape and size of the target organ, the blood flow of the target organ, the concentration of the drug at the target site, the expression level of enzymes, the plasma protein binding rate, the volume of distribution, and the clearance rate, are data that are difficult to truly sense in the physical entity.

[0121] Service data includes historical data, statistical data, and medical records, especially physical and virtual entity data related to adult physiological pharmacokinetics. Knowledge data includes expert knowledge, treatment guidelines, rule constraints, and commonly used algorithm libraries. Fusion-derived data is derived from the above-mentioned physical and virtual entity data on elderly physiological pharmacokinetics, service data, and knowledge data through data transformation, preprocessing, classification, association, integration, and fusion. By fusing physical reality data with multi-temporal and spatial related data, historical statistical data, expert knowledge, and other information data, cyber-physical fusion data is obtained, thus reflecting more comprehensive and accurate information and realizing information sharing and value-added. Fusion-derived data includes guiding dosage data for elderly medication, judgmental data on actual medication results obtained by the physiological pharmacokinetics physical entity data acquisition module, judgmental data on the virtual operation results of the elderly physiological pharmacokinetics virtual entity module, data that self-updates the twin data itself, and data that updates the "elderly physiological pharmacokinetics virtual entity".

[0122] The following is combined Figure 15 The diagram shown is a schematic of the application system of a digital twin model of physiological pharmacokinetics in the elderly. Figure 6 The "Flowchart for Application of Physiological Pharmacokinetic Digital Twin Model" shown describes how to use the physiological pharmacokinetics digital twin model of the elderly described in this invention to determine the dosage of a certain drug in the elderly body.

[0123] Figure 6 In the first step, based on the adult physiological pharmacokinetic physical entity data and virtual entity data in the "service data" of the twin data, the initial dose of a specific elderly patient is extrapolated by fusing derived data, and then... Figure 15 The information flow shown in ② is to input the initial dose to the "virtual entity of physiological pharmacokinetics of the elderly".

[0124] The second step involves the creation of a "virtual entity for the physiological pharmacokinetics of the elderly" through... Figure 15 After receiving the initial dose, the information flow ② shown in the diagram first performs virtual operation, through... Figure 15 The information flow shown in ③ ensures the interaction between the elderly body model and the drug model, generates operational results, and transmits various key observation data through... Figure 15 The information flow shown in ④ is output to the "virtual entity data of elderly physiological pharmacokinetics" in the twin data, especially the drug concentration data at the target site, plasma protein binding rate, liver clearance rate, glomerular filtration rate, creatinine clearance rate and other data that cannot be sensed in the "physical entity data acquisition module of elderly physiological pharmacokinetics".

[0125] The third step involves using the twin data's "virtual entity data on the physiological pharmacokinetics of the elderly" through... Figure 15 The information flow shown in ④, after obtaining the data from the "virtual entity of physiological pharmacokinetics in the elderly," focuses on characteristic data such as target site physiological changes, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters. This data is then combined with service and knowledge data from the twin database and utilized... Figure 15 The information flow shown in section ⑤ employs data mining and analysis techniques to extract useful information from the original data. Based on expert knowledge, rule base information, and other data, it generates fused derived data, outputting the effect of the initial dose virtual run. The effect of the virtual run determines whether dose adjustment and re-run are necessary. If adjustment and re-run are required, the fused derived data also outputs a guiding dose for the next virtual run. Specifically, neural network analysis, dynamic Bayesian networks, and Dynamic Data Driven Analysis and Decision Support (DDDAS) techniques can be used to fuse the derived data. More specifically, in this embodiment, the hidden layer of the neural network is built from service data and knowledge data and generated through multiple iterative calculations. The aforementioned physiological changes at the target site in the elderly, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters are used as the input layer data (X1, X2...X...X...X...) of the neural network. n After processing by a neural network algorithm, the output layer can output pharmacokinetic data, pharmacodynamic data, and guidance doses for the next administration (O1, O2...O...). n The fusion-derived data will be used to determine whether a new virtual run is needed, based on the algorithm-generated guideline dosage for the next medication.

[0126] Fourth, if the results of the fusion and derived data determine that the previous virtual run of the "virtual entity of physiological pharmacokinetics in the elderly" needs to be re-run with adjusted dosage, then... Figure 15 The information flow shown in step ② re-outputs the guidance dose to the "virtual entity of physiological pharmacokinetics in the elderly," and then repeats steps two and three above. If the fusion-derived data results determine that the previous virtual run of the "virtual entity of physiological pharmacokinetics in the elderly" did not require dose adjustment, then... Figure 15 The information flow shown in section ① provides the confirmed dose to the output module, which is used by the doctor to administer this medication to the elderly patient. Figure 15 The information flow shown in section ② inputs the confirmed dose into the "virtual entity of physiological pharmacokinetics of the elderly".

[0127] The fifth step involves information flow ① obtaining the medication dosage and administering the medication. After elderly patients receive the medication, their bodies produce corresponding bodily responses, represented by physiological changes, changes in target site function, and changes in blood drug concentration. Data is then transmitted through sensing to obtain data from medical examinations and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, or biochemical data such as complete blood counts, urinalysis, and enzyme tests. All of this data is then processed by the elderly physiological pharmacokinetics physical data acquisition module. Figure 15 The information flow shown in diagram ⑥ is provided to the physical entity data of elderly physiological pharmacokinetics. Simultaneously, the virtual entity of elderly physiological pharmacokinetics also runs virtually, generating corresponding results, and... Figure 15 The information flow shown in ④ is provided to the virtual entity data on the physiological pharmacokinetics of the elderly.

[0128] Step 6, twin data through Figure 15 The information flows shown in ⑥ and ④, respectively, yield the data collected in step 5 through the physical entity data acquisition module for elderly physiological pharmacokinetics and the data from the virtual entity operation for elderly physiological pharmacokinetics. The key focus is on comparing the differences between the physical entity data and the virtual entity data for elderly physiological pharmacokinetics. This is then combined with service data and knowledge data from the twin data, utilizing... Figure 15 The information flow shown in section ⑤ undergoes data fusion processing, and then, combined with expert knowledge, rule base information, and other data, fusion-derived data is generated to produce result determination data for the previous medication dosage in elderly patients. On the other hand, based on the fusion-derived data, through... Figure 15 The information flow shown in section ⑦ corrects the virtual entity of physiological pharmacokinetics of the elderly, so that the virtual entity of physiological pharmacokinetics of the elderly can achieve the effect of synchronous evolution with the elderly patients.

[0129] Step 7: If the data collected by the elderly physiological pharmacokinetic physical entity data acquisition module indicates that the elderly patient has achieved the expected drug efficacy and can discontinue medication, or if the relevant physiological and biochemical indicators have reached the threshold for necessary discontinuation, then medication is discontinued. If continued medication is required, the fusion-derived data will output the guiding dose for the next administration, and steps 2 through 6 above will be repeated.

[0130] By adopting the aforementioned methods for establishing and applying digital twin models of pharmacokinetic physiology in the elderly, the safety of medication use in the elderly can be guaranteed to the greatest extent possible, and precise control of dosage can be achieved for each administration. Furthermore, if a personalized virtual entity and twin data of pharmacokinetic physiology can be established for each specific elderly individual, this personalized virtual entity and twin data can grow and evolve alongside that individual. This not only allows the results of virtual operation to more realistically reflect the elderly patient's condition but also provides greater support for healthcare services throughout their lifespan.

[0131] Example 4: Digital Twin Dosage Assessment System for Pregnant Women

[0132] The safety of medication use during pregnancy for both mother and fetus has always been a concern for doctors and pregnant women. Most drugs can cross the placenta and enter the fetus, affecting normal fetal development and, in severe cases, causing birth defects and stillbirth. Pregnant women inevitably use medication during pregnancy; statistics show that approximately 80% of pregnant women use medication during this period. Early pregnancy is a stage of differentiation for the fetus's various parts and organs, and drug-induced birth defects are more likely to occur during this stage. Pregnant women must pay particular attention to medication safety during this period. Determining a safe and effective starting dose for pregnant women has always been a challenge in drug development trials. During pregnancy, to adapt to the needs of fetal growth and development, a series of physiological changes occur in various systems of the pregnant woman's body, resulting in changes in drug metabolism pharmacokinetics. Drug absorption is slower in the mother during pregnancy, and the peak absorption is delayed and often lower. Blood volume increases by approximately 35%-50% during pregnancy, resulting in lower blood drug concentrations in pregnant women compared to non-pregnant women. Drug protein binding capacity decreases, the rate of drug clearance from the liver slows down, and the rate of drug excretion from the kidneys increases. Drugs can be transported and metabolized across the placenta. Because the placental barrier cannot completely protect the fetus from the effects of drugs, most drugs can cross the placenta into the fetal body, and a considerable number of drugs can form harmful substances after metabolism, leading to embryonic death or malformation. Drugs in amniotic fluid can be swallowed by the fetus and enter the gastrointestinal tract, and then absorbed into the fetal bloodstream. Their metabolites are excreted in fetal urine, and some of these excreted metabolites can be reabsorbed into the fetal bloodstream, forming the amniotic fluid intestinal circulation. Furthermore, the blood volume has a significant impact on the distribution of drugs in the fetus; the distribution of drugs in the fetus is consistent with the distribution in the fetal bloodstream. Because the fetus's liver, brain, and other organs are relatively large in proportion to its body, and have high blood flow, drugs tend to concentrate in the liver and brain. Many drugs are metabolized in the liver; due to the lack of enzymes in the fetal liver, the ability to metabolize drugs is low, resulting in higher concentrations of certain drugs in the fetal blood than in the mother. From 11-14 weeks of gestation, although the fetal kidneys have excretory function, the excretion of drugs and their degradation products is delayed due to the low glomerular filtration rate, and the excretory capacity of the biliary tract is also weak. Furthermore, after metabolism, the drug forms substances with high polarity and water solubility, making it difficult to cross the placental barrier and be transported to the mother, leading to accumulation and damage in the body. Simultaneously, because the drug may be re-swallowed after being excreted into the amniotic cavity, a "membranous fluid intestinal circulation" occurs. Before administering medication to pregnant women, the pharmacokinetic characteristics between the mother, placenta, and fetus, as well as the pharmacological effects of the drug across the placenta on the embryo, fetus, and newborn, must be fully considered. Currently, the administration method for pregnant women is usually inferred from the non-pregnant state. This results in a higher risk of sub-therapeutic effects and toxic side effects for the mother and / or fetus.

[0133] This embodiment employs a method for establishing and applying a digital twin model of pregnant women's physiological pharmacokinetics, and proposes a method for determining the optimal medication dosage for pregnant women based on the digital twin model. The overall structure of the digital twin model of pregnant women's physiological pharmacokinetics comprises five main parts: a module for inputting basic information of patients with hepatic or renal insufficiency, a module for acquiring physical entity data of pregnant women's physiological pharmacokinetics, a virtual entity of pregnant women's physiological pharmacokinetics, a twin data module, and an output module. The physical entity data acquisition module of pregnant women's physiological pharmacokinetics can obtain data through sensing devices, map it to the virtual entity module of pregnant women's physiological pharmacokinetics, and update the twin data. The virtual entity module of pregnant women's physiological pharmacokinetics and the twin data module achieve virtual operation and mutual optimization and iteration under interactive conditions. The three parts—the physical entity data acquisition module of pregnant women's physiological pharmacokinetics, the virtual entity module of pregnant women's physiological pharmacokinetics, and the twin data module—are interconnected, mutually optimizing and iteratively updating, and evolving synchronously.

[0134] The physical pharmacokinetic data acquisition module for pregnant women collects physical pharmacokinetic data from medical testing equipment after drug administration. This data includes blood drug concentration, CT scans, MRI, electrocardiograms, color Doppler ultrasound, complete blood counts, urinalysis, and bioenzyme data. This physical pharmacokinetic data is then transmitted to the twin data module. The unit level, also known as the tissue level, includes epithelial tissue, connective tissue, muscle tissue, and nervous tissue. Above the unit level is the subsystem level, also known as the organ level, which includes organs such as the heart, brain, liver, lungs, kidneys, gonads, mammary glands, uterus, placenta, amniotic fluid, fetal heart, fetal brain, fetal liver, and fetal kidneys. Above the subsystem level is the system level, which includes the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, reproductive system, and fetus. The above three levels of systems constitute an organism, which is organically integrated with each other. During its operation, data such as body temperature, blood pressure, blood oxygen, electrocardiogram data, imaging data of some organs, laboratory data, height, weight, gestational age, and fetal size can be obtained through certain equipment. However, most data that characterize the health status of various tissues and the status of the system, such as plasma protein binding rate, gastrointestinal physiological parameters, liver clearance rate parameters, glomerular filtration rate, creatinine clearance rate, fetal blood drug concentration, and amniotic fluid drug residue concentration, cannot be directly measured. These parameters are the ones that pharmacokinetics is more concerned with. This requires the establishment of a virtual entity of pregnant women's physiological pharmacokinetics and the direct acquisition of virtual parameters from the model.

[0135] The virtual entity module for the physiological pharmacokinetics of pregnant women comprises two main models: a pregnant woman's body model and a drug model. The pregnant woman's body model includes nine parts: maternal anatomical models, tissue models, physiological models, biochemical models, fetal anatomical models, tissue models, physiological models, biochemical models, and a placental model. The drug model includes four parts: physical models, chemical models, toxicological models, and pharmacological models. The "virtual entity for the physiological pharmacokinetics of pregnant women" must continuously iterate and grow through processes such as model construction, model assembly, model fusion, model validation, model updating, and model management. The model updating process is particularly crucial. It refers to updating the model when there are discrepancies between the results of the virtual model operation and the physical objects generated by the physical entity operation, or when the model fails to meet requirements, to make the model more closely approximate the actual state or characteristics of the physical objects.

[0136] The twin data module includes physical entity data on the physiological pharmacokinetics of pregnant women, virtual entity data on the physiological pharmacokinetics of pregnant women, service data, knowledge data, and fused derived data. The physical entity data on the physiological pharmacokinetics of pregnant women refers to data that reflects the physical function and state of the pregnant woman, as well as the working conditions of its internal tissues, organs, and systems. This data comes from sensor data of the entity, including data from medical tests and scanning instruments such as CT scans, MRI, electrocardiograms, and color Doppler ultrasound, and biochemical data such as blood routine tests, urine tests, and biological enzymes. In the twin data of this invention, the focus is on data related to the pharmacokinetic characteristics of drugs in the human body, such as data on absorption, distribution, metabolism, and excretion. Only a small portion of this data is actually measurable, such as height and weight, drug concentration in the pregnant woman's blood, and blood flow velocity; other data are difficult to measure.

[0137] The virtual entity data on the physiological pharmacokinetics of pregnant women does not rely on the data obtained from the physical entity data acquisition module for the physiological pharmacokinetics of pregnant and elderly women. Instead, it directly obtains data from the mathematical model within the "virtual entity of physiological pharmacokinetics of pregnant women." This includes various data generated by the virtual operation of the pregnant body model and the drug model, as well as data generated by the interaction between the drug model and the pregnant body model. In particular, it includes data directly related to pharmacokinetic characteristics, such as the shape and size of the target organ, the blood flow of the target organ, the concentration of the drug at the target site, the expression level of enzymes, the plasma protein binding rate, the volume of distribution, the clearance rate, placental metabolism, and fetal blood drug concentration, which are difficult to truly sense in the physical entity.

[0138] Service data includes historical data, statistical data, and medical records, especially physical and virtual entity data related to adult physiological pharmacokinetics. Knowledge data includes expert knowledge, medical industry standards, rule constraints, and commonly used algorithm libraries. Fusion-derived data is derived from the above-mentioned physical and virtual entity data, service data, and knowledge data of pregnant women through data transformation, preprocessing, classification, association, integration, and fusion. By fusing physical reality data with multi-temporal and spatial related data, historical statistical data, expert knowledge, and other information data, cyber-physical fusion data is obtained, thereby reflecting more comprehensive and accurate information and realizing information sharing and value-added. Fusion-derived data includes guiding dosage data for medication use in pregnant women, judgmental data on actual medication results in pregnant women, judgmental data on the virtual operation results of the virtual entity of physiological pharmacokinetics in pregnant women, data that self-updates the twin data itself, and data that updates the virtual entity of physiological pharmacokinetics in pregnant women.

[0139] Figure 19 A flowchart was created for a virtual entity model of pregnant women's physiological pharmacokinetics. During pregnancy, the embryo rapidly develops from a single cell into an individual, during which both the mother and fetus undergo significant changes. A virtual entity model of pregnant women's physiological pharmacokinetics was established based on embryological theoretical data and its virtual operation was implemented. The virtual entity was then continuously updated based on the pregnant woman's regular prenatal checkup data to ensure that the virtual entity's physiological pharmacokinetics development was synchronized with the pregnant woman's.

[0140] The following is combined Figure 20 The diagram shown is a "Digital Twin Model Application System for Pregnant Women's Physiological Pharmacokinetics". Figure 6 The "Digital Twin Model Application Flowchart" shown describes how to use the digital twin model of pregnant women's physiological pharmacokinetics described in this invention to determine the dosage of a certain drug in a pregnant woman's body.

[0141] Figure 6 In the first step, based on the adult physiological pharmacokinetic physical entity data and virtual entity data in the "service data" of the twin data, the initial dose of a specific drug for pregnant women is extrapolated by fusing derived data, and then... Figure 20 The information flow shown in section ② inputs the initial dose to the "virtual entity of pregnant woman's physiological pharmacokinetics".

[0142] The second step involves a virtual entity representing the physiological pharmacokinetics of pregnant women. Figure 20 After receiving the initial dose, the information flow ② shown in the diagram first performs virtual operation, through... Figure 20 The information flow shown in ③ ensures the interaction between the pregnant woman's body model and the drug model, generates operational results, and transmits various key observation data through... Figure 20The information flow shown in ④ is output to the "virtual entity data of pregnant women's physiological pharmacokinetics" in the twin data, especially the drug concentration data at the target site, plasma protein binding rate, liver clearance rate, glomerular filtration rate, creatinine clearance rate, fetal blood drug concentration, amniotic fluid drug residue concentration and other data that cannot be sensed in the physical entity data acquisition module of pregnant women's physiological pharmacokinetics.

[0143] The third step involves using the twin data's "virtual entity data on the physiological pharmacokinetics of pregnant women" through... Figure 20 As shown in information flow ④, after obtaining the data from the "Virtual Entity of Pregnant Women's Physiological Pharmacokinetics," the focus is on characteristic data such as physiological changes at the target site in pregnant women, protein binding rate in pregnant women, gastrointestinal physiological parameters, liver clearance rate, renal function parameters, fetal blood drug concentration, and fetal drug clearance rate. This data is then combined with service and knowledge data from the twin database and utilized... Figure 20 The information flow shown in section ⑤ employs data mining and analysis techniques to extract useful information from the original data. Based on expert knowledge, rule base information, and other data, it generates fused derived data, outputting the effect of the initial dose virtual run. The effect of the virtual run determines whether dose adjustment and re-running are necessary. If dose adjustment and re-running are required, the fused derived data also outputs the guiding dose for the next virtual run. Specifically, neural network analysis, dynamic Bayesian networks, and Dynamic Data Driven Analysis and Decision Support (DDDAS) techniques can be used to fuse the derived data. More specifically, in this embodiment, the hidden layer of the neural network is built from service data and knowledge data and generated through multiple iterative calculations. The aforementioned physiological changes at the target site in pregnant women, protein binding rate, gastrointestinal physiological parameters, liver clearance rate, and renal function parameters are used as the input layer data (X1, X2...X...X...X...) of the neural network. n After processing by a neural network algorithm, the output layer can output pharmacokinetic data, pharmacodynamic data, and guidance doses for the next administration (O1, O2...O...). n The fusion-derived data will be used to determine whether a new virtual run is needed, based on the algorithm-generated guideline dosage for the next medication.

[0144] Fourth, if the results of the fusion and derived data indicate that the previous virtual run of the "virtual entity of pregnant women's physiological pharmacokinetics" needs to be re-run with dosage adjustments, then... Figure 20 The information flow shown in step ② re-outputs the guidance dose to the "virtual entity of pregnant women's physiological pharmacokinetics," and then repeats steps two and three above. If the fusion-derived data results determine that the previous virtual run of the "virtual entity of pregnant women's physiological pharmacokinetics" did not require dose adjustment, then... Figure 20 The information flow shown in section ① will provide the confirmed dosage to the pregnant woman, while simultaneously... Figure 20The information flow shown in section ② inputs the confirmed dose into the pregnant woman's physiological pharmacokinetics virtual entity.

[0145] The fifth step involves the pregnant woman receiving the prescribed dosage and undergoing medication. After administration, the body produces corresponding physiological responses, including changes in target site function and blood drug concentration. Data is then transmitted through sensors, yielding data from medical examinations and scanning instruments such as CT scans, MRIs, electrocardiograms, and color Doppler ultrasounds, as well as biochemical data such as complete blood counts, urinalysis, and enzyme tests. This data is then processed... Figure 20 The information flow shown in diagram ⑥ is provided to the physical entity data of the pregnant woman's physiological pharmacokinetics. Simultaneously, the virtual entity of the pregnant woman's physiological pharmacokinetics also runs virtually, generating corresponding results, and... Figure 20 The information flow shown in ④ is provided to the virtual entity data on the physiological pharmacokinetics of pregnant women.

[0146] Step 6, twin data through Figure 20 The information flows shown in ⑥ and ④, after obtaining the data from the physical entity data acquisition module and the virtual entity data of the pregnant woman's physiological pharmacokinetics generated in step 5, respectively, focus on comparing the differences between the physical entity data and the virtual entity data of the pregnant woman's physiological pharmacokinetics. Then, combining this with the service data and knowledge data in the twin data, and utilizing... Figure 20 The information flow shown in section ⑤ undergoes data fusion processing, and then, combined with expert knowledge, rule base information, and other data, fusion-derived data is generated to produce physiological pharmacokinetic data of the previous drug dosage in the pregnant woman. On the other hand, based on the fusion-derived data, through... Figure 20 The information flow shown in section ⑦ corrects the virtual entity of the pregnant woman's physiological pharmacokinetics, so that the virtual entity of the pregnant woman's physiological pharmacokinetics can achieve the effect of synchronous evolution with the pregnant woman who is being treated.

[0147] Step 7: If the previous medication dosage was collected from the pregnant woman's physiological pharmacokinetic physical entity data acquisition module, and it is determined that the pregnant woman who received the previous medication has achieved the expected drug efficacy and can discontinue the medication, or if the relevant physiological and biochemical indicators have reached the threshold for necessary discontinuation of the medication, then the medication is discontinued. If continued medication is required, the fusion-derived data will output the guiding dosage for the next medication, and steps 2 through 6 above will be repeated.

[0148] By adopting the aforementioned methods for establishing and applying the digital twin model of pregnant women's physiological pharmacokinetics, the safety of medication use during pregnancy can be maximized, and precise control of dosage can be achieved for each administration. Furthermore, if a personalized "virtual entity of pregnant women's physiological pharmacokinetics" and "twin data" can be established for each individual pregnant woman, these personalized entities can grow and evolve alongside that specific individual. This not only allows the virtual operation results to more realistically reflect the pregnant woman's condition but also provides more comprehensive medical service support throughout the pregnancy.

Claims

1. A drug administration and dosage assessment system based on digital twins, characterized in that, The system includes the following modules: a basic information input module for drug recipients, a physical entity data acquisition module for physiological pharmacokinetics, a virtual entity module for physiological pharmacokinetics, a twin data module, and an output module. The twin data module is connected to each of these modules. The basic information input module collects basic information about the drug recipients and transmits this information to the twin data module. The physical entity data acquisition module collects pharmacokinetic data from the drug recipients and transmits this data to the twin data module. The virtual entity module digitally simulates the drug recipients, generating pharmacokinetic data through model simulation and transmitting the generated data to the twin data module. The twin data module performs fusion processing on the received data, determines whether to discontinue medication or provides a guide dose for the next administration, and outputs the discontinuation decision or the guide dose for the next administration to the output module. The twin data module includes: physical entity data of physiological pharmacokinetics, virtual entity data of physiological pharmacokinetics, service data, knowledge data, and fusion-derived data sub-modules. Virtual entity pharmacokinetics data obtained by the virtual entity module through virtual operation is stored in the virtual entity data sub-module. Physical entity pharmacokinetics data collected by the physical entity data acquisition module is stored in the physical entity data sub-module. The service data sub-module includes adult physiological pharmacokinetics physical entity data and virtual entity data. The knowledge data sub-module includes expert knowledge, medical industry standards, treatment guidelines, and commonly used algorithm libraries. The fusion-derived data sub-module is derived data obtained by processing the aforementioned physical entity data of physiological pharmacokinetics, virtual entity data of physiological pharmacokinetics, service data, and knowledge data.

2. The digital twin-based drug delivery and dosage assessment system according to claim 1, characterized in that, For first-time medication use, the dosage assessment system outputs the prescribed dosage using the following method: (1) The twin data module generates the first dose of the drug administration subject based on the basic information of the drug administration subject, and transmits the first dose data to the physiological pharmacokinetic virtual entity module; (2) The virtual entity module selects or creates a virtual entity model that is appropriate to the basic information of the drug administration object and performs virtual calculations. (3) The twin data module evaluates the effect of the drug dosage simulation operation based on the data generated by the virtual operation of the physiological pharmacokinetics virtual entity module, and determines whether the dosage needs to be adjusted. If the dosage needs to be adjusted, the twin data module adjusts the dosage and sends the dosage back to the physiological pharmacokinetics virtual entity module to repeat steps (2)-(3) until the dosage does not need to be adjusted. Then the dosage is pushed to the output module as the first drug guidance dosage.

3. The digital twin-based drug delivery and dosage assessment system according to claim 1, characterized in that, For non-first-time medication use, the dosage assessment system outputs medication guidance dosage through the following method: a. After the drug is administered to the recipient, the physiological pharmacokinetic physical entity data acquisition module collects the pharmacokinetic data of the recipient and transmits the collected data to the twin data module; b. The physiological pharmacokinetic virtual entity performs virtual operation based on the drug dosage of the drug recipient, and transmits the obtained data to the twin data module; c. The twin data module corrects the pharmacokinetic data of the obtained physiological pharmacokinetic virtual entity based on the pharmacokinetic data of its target drug, and also corrects the physiological pharmacokinetic virtual entity module; d. The twin data module determines the effect of the previous drug dose on the drug recipient based on the pharmacokinetic data of the physical entity and the pharmacokinetic data of the virtual entity obtained from the operation. If the effect achieves the expected drug efficacy, or if the relevant physiological and biochemical indicators reach the threshold that must be stopped, the drug discontinuation instruction will be output to the output module. If the drug needs to be continued, the guidance dose for the next drug administration will be output to the output module.

4. The drug delivery and dosage assessment system based on digital twins according to claim 1, characterized in that, The physiological pharmacokinetic physical entity data acquisition module obtains pharmacokinetic data of the physical entity of the drug recipient through medical testing equipment after drug administration. The physical entity pharmacokinetic data includes blood drug concentration, CT, MRI, electrocardiogram, color Doppler ultrasound, blood routine, urine test, and biological enzyme data, and transmits the physical entity pharmacokinetic data to the twin data module.

5. The digital twin-based drug delivery and dosage assessment system according to claim 1, characterized in that, The physiological pharmacokinetic physical entity data acquisition module includes three levels of pharmacokinetic data acquisition sub-modules: unit level, subsystem level, and system level. The unit level includes medical data acquisition of at least one tissue, such as epithelial tissue, connective tissue, muscle tissue, and nerve tissue; the subsystem level includes medical data acquisition of at least one organ, such as the heart, brain, liver, lungs, and kidneys; and the system level includes medical data acquisition of at least one system, such as the digestive system, respiratory system, circulatory system, endocrine system, nervous system, musculoskeletal system, urinary system, and reproductive system.

6. The digital twin-based drug delivery and dosage assessment system according to claim 1, characterized in that, The physiological pharmacokinetics virtual entity module includes two main models: an organism model and a drug model. The organism model includes four sub-models: an anatomical model, a tissue model, a physiological model, and a biochemical model. The drug model includes four sub-models: a physical model, a chemical model, a toxicological model, and a pharmacological model.

7. A method for evaluating drug dosage using the system according to any one of claims 1-6, characterized in that, Includes the following steps: (1) The twin data module generates the first dose based on the basic information of the drug recipient and pushes the first dose to the physiological pharmacokinetic virtual entity module; (2) The physiological pharmacokinetics virtual entity module performs virtual operation based on the first dose or dose pushed by the twin data module to generate operation results and pushes the data to the twin data module; (3) The twin data module evaluates the effect of the drug dosage simulation operation based on the data generated by the virtual operation results; determines whether the dosage needs to be adjusted. If the dosage needs to be adjusted, the twin data module adjusts the dosage and repeats steps (2)-(3). If no dose adjustment is required, the drug dose is pushed to the output module to administer the drug to the recipient, and at the same time the drug dose is transmitted to the physiological pharmacokinetics virtual entity module; (4) After the drug is administered to the drug recipient, the physiological pharmacokinetic physical entity data acquisition module collects the pharmacokinetic data of the drug recipient, obtains the physical entity pharmacokinetic data, and transmits it to the physical entity pharmacokinetic data in the twin data module; at the same time, the physiological pharmacokinetic virtual entity performs virtual operation according to the drug dosage pushed in step (3), and transmits the data obtained from the operation to the virtual entity pharmacokinetic data in the twin data module. (5) The twin data module corrects the pharmacokinetic data of the physiological pharmacokinetic virtual entity based on its physical entity pharmacokinetic data, and also corrects the physiological pharmacokinetic virtual entity module. (6) The twin data module judges the effect of the previous drug dose on the drug recipient based on the pharmacokinetic data of the physical entity and the pharmacokinetic data of the virtual entity. If the effect reaches the expected drug efficacy, or the relevant physiological and biochemical indicators reach the limit that must be stopped, the drug is stopped and the drug stop instruction is output to the output module. If it is necessary to continue the drug or adjust the dose, the guidance dose for the next drug is output to the output module, and then steps (2)-(6) are repeated.

8. The method for evaluating drug dosage according to claim 7, characterized in that, In step (1), the initial drug dose is obtained based on the service data submodule in the twin data module; in step (2), virtual entity physiological pharmacokinetic data is obtained through the interaction between the organism model and the drug model, the virtual entity physiological pharmacokinetic data including drug concentration data at the target site, drug efficacy data, and toxicity in key areas of the heart and brain; in step (3), the twin data module determines the effect of the initial dose simulation by combining its service data and knowledge data submodules, and then determines whether the dose needs to be adjusted and the simulation run is repeated through the generated fusion-derived data; in step (5), the twin data module performs data fusion processing by combining its service data and knowledge data submodules, and corrects the physiological pharmacokinetic virtual entity module according to the data processing results; in step (6), the twin data module determines the effect of the drug through the data fusion processing results in step (5).

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a data processing program, which, when executed by a processor, implements the method for evaluating a drug dosage as described in any one of claims 7-8.

Citation Information

Patent Citations

  • Clinical safe and reasonable medication evaluation method and system based on digital twinning

    CN114758795A

  • Systems and methods for predicting and adjusting the dosage of medicines in individual patients

    US20160335412A1