A method for precise administration of oxytocin injection based on PBPK model optimization

Abstract

This invention provides a precise dosing method for oxytocin injection based on a physiological pharmacokinetic (PBPK) model, belonging to the field of personalized drug delivery technology. The method first collects individual physiological parameter information of the target receptor, constructs a personalized PBPK model, and performs simulation calculations based on the drug characteristics of oxytocin to predict the in vivo drug distribution under different dosing regimens. By evaluating the drug concentration, dosing duration, and predicted uterine contraction response of each regimen in uterine tissue and systemic circulation, the optimal combination of dosing parameters that simultaneously meets the requirements of efficacy and safety is selected. Based on this combination, injection control instructions are generated, and precise infusion is implemented through a syringe system with programmable control functionality. This method achieves model-driven personalized precision control, effectively improving the safety and clinical efficacy of oxytocin administration.

Description

Technical Field

[0001] This invention relates to the fields of drug delivery and personalized medicine, and in particular to a precise delivery method for oxytocin injection based on a PBPK model optimization. Background Technology

[0002] In existing technologies, oxytocin, as a commonly used uterine contraction agent, is widely used in clinical scenarios such as labor induction, labor stimulation, and postpartum hemostasis. The administration method of its injection solution is usually based on clinical experience, including a fixed initial rate, a timed incremental drip rate, or manual adjustment of the infusion rate based on uterine contraction responses. Some automated injection devices have drip rate control functions, but the control logic is mostly based on preset rules or limited parameter feedback, failing to fully reflect individual physiological differences.

[0003] However, existing technologies still have significant shortcomings in practical applications. On the one hand, fixed or standardized dosing patterns cannot be dynamically adapted to key factors such as individual weight, liver and kidney function, or placental blood flow, which can easily lead to insufficient or excessive uterine contractions, thereby increasing the risk of induction failure or inducing excessive uterine contractions. On the other hand, most existing automated injection devices are not linked to pharmacokinetic models and lack mechanisms for predicting and regulating drug concentration distribution, limiting the feasibility of personalized precision medication.

[0004] To overcome the above problems, there is an urgent need to provide an optimized method for oxytocin administration that can integrate individual physiological characteristics and has predictive and regulatory capabilities. Summary of the Invention

[0005] This application provides a precise oxytocin injection method optimized based on the PBPK model, which realizes precise oxytocin administration based on individual physiological characteristics, and significantly improves the effectiveness and safety of medication.

[0006] This application provides a method for precise administration of oxytocin injection based on a PBPK model optimization, including:

[0007] Collect individual physiological parameter information of the target receptor;

[0008] Based on the individualized parameter dataset, a physiological pharmacokinetic model suitable for the target receptor was constructed, and the physicochemical properties, tissue distribution parameters, metabolic and excretion parameters of the drug related to oxytocin injection were set to simulate the in vivo concentration change process under different administration conditions.

[0009] Multiple sets of drug administration parameters, including injection initiation rate, maintenance rate and duration of administration, are input into the physiological pharmacokinetic model. Simulation calculations are performed to obtain the corresponding concentration change curves. Based on these curves, the drug concentration level, duration of administration and predicted uterine contraction response of each set of drug administration parameters in uterine tissue and systemic circulation are evaluated.

[0010] The target dosing parameter combination that simultaneously meets the following clinical control conditions is selected from the simulation results: the drug concentration in the uterine tissue reaches the set effective concentration, the drug concentration in the systemic circulation does not exceed the set safety limit, the dosing duration is within the set range, and the predicted value of uterine contraction response is within the target range.

[0011] Based on the selected target dosing parameter combination, an injection control command is generated, which includes the injection initiation rate, variable speed adjustment information, maintenance rate and termination time, and is converted into a format that can be recognized by the program-controlled injection device.

[0012] The injection control command is imported into a syringe system with program control function, and the precise administration of oxytocin injection is performed.

[0013] The beneficial effects of the technical solution provided in this application include:

[0014] (1) By collecting individual physiological parameters such as the target recipient's weight, liver and kidney function, and placental blood flow and constructing an individualized PBPK model, the dosing strategy can be dynamically adjusted according to individual differences, significantly improving the adaptability and effectiveness of oxytocin administration. (2) During the simulation evaluation process, the uterine drug concentration, systemic blood drug concentration, and dosing duration are comprehensively limited to ensure that the drug can reach the effective concentration without exceeding the safety threshold, reducing the risk of complications such as excessive uterine contractions and fetal distress. (3) By predicting the drug response and screening the optimal combination of dosing parameters through the PBPK model, the programmed control instructions can be generated and executed directly without the need for repeated manual trials of dosage and rate, which can effectively shorten the labor initiation time and reduce the workload of clinical regulation. (4) The injection control instructions can be imported into a syringe system with programmable control function to realize the automatic control of the entire process from parameter modeling and scheme optimization to injection execution, improve the stability and consistency of drug infusion, and promote the transformation of uterine contraction agent use from experience-based dosing to model-driven. Attached Figure Description

[0015] Figure 1 This is a flowchart of a precise oxytocin injection method optimized based on a PBPK model, provided in the first embodiment of this application. Detailed Implementation

[0016] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0017] The first embodiment of this application provides a precise drug delivery method for oxytocin injection based on a PBPK model optimization. Please refer to... Figure 1 This figure is a schematic diagram of the first embodiment of this application. The following is in conjunction with... Figure 1 The first embodiment of this application provides a detailed description of a precise oxytocin injection method optimized based on a PBPK model.

[0018] Step S101: Collect individual physiological parameter information of the target recipient, including weight, height, age, liver function, kidney function, reproductive system status and placental blood flow, and form an individualized parameter dataset.

[0019] In step S101, the system needs to comprehensively collect individual physiological parameter information of the target receptor and standardize and organize it into a dataset that can be used for model calculation. This step is the foundation for all subsequent model construction and drug delivery optimization calculations, and must have high accuracy, completeness, and uniformity in format. Any omissions or errors may affect the final pharmacokinetic prediction results.

[0020] Specifically, during implementation, the recipient's basic vital signs should first be obtained through clinical data collection devices or electronic medical record systems, including but not limited to weight (kg), height (cm), and age (years). Weight and height will be used to estimate model structural parameters such as body surface area and tissue / organ volume distribution, while age serves as a key variable regulating metabolic function and organ maturity.

[0021] Secondly, liver and kidney function indicators of the target receptor need to be obtained. Liver function is mainly evaluated through clinical indicators such as serum transaminases (e.g., ALT, AST), albumin levels, and bilirubin concentration, reflecting the liver's ability to metabolize oxytocin. Kidney function is estimated using parameters such as serum creatinine, blood urea nitrogen, and creatinine clearance rate to estimate the glomerular filtration rate (GFR) in the drug clearance pathway. These metabolic and excretion variables should be input into the system in quantitative form and updated according to the current physical examination time point to ensure that the model calculations are representative in real time.

[0022] In addition, the recipient's reproductive system status needs to be assessed, particularly the sensitive stage of the uterus and the gestational period. During implementation, data such as ultrasound reports related to uterine blood perfusion, reproductive hormone levels (e.g., estrogen, progesterone), and uterine contraction monitoring curves should be obtained to determine the uterine receptor's response threshold to oxytocin and its potential activity level. This information is primarily used to predict the intensity of oxytocin's pharmacodynamic response in local uterine tissues and to estimate parameters affecting uterine contraction responses in the model.

[0023] A more critical parameter is placental blood flow, which directly affects the distribution and transport of oxytocin between the mother and placenta. Data acquisition methods typically include placental Doppler ultrasound to obtain parameters such as uterine artery, umbilical artery, and venous flow velocity, pulsatility index (PI), and resistance index (RI) to reflect placental perfusion efficiency and blood exchange capacity. These parameters should be converted into standardized input variables and correlated with a uterine tissue blood flow proportion model to calculate the local peak concentration of oxytocin in the uterine receptor region.

[0024] After all the above physiological parameter information has been collected, it should undergo data cleaning, unit standardization, and missing value imputation to construct a complete and structured individualized parameter dataset. This dataset should adopt a tabular structure, with field names consistent with the variable names required in the physiological pharmacokinetic model, and should support direct interface with the model input interface. The data should include parameter names, values, units, measurement time points, and data source records to ensure traceability and accuracy.

[0025] To ensure the clinical feasibility of data collection, it is recommended to develop an automatic data acquisition interface based on the integration of hospital information systems and laboratory information systems. This interface should associate and match the target patient's ID with the corresponding physiological data, enabling one-click import. Simultaneously, for indicators that cannot be automatically collected (such as placental blood perfusion parameters), supplementary data should be obtained through clinical ultrasound equipment or manual entry, with necessary data validation logic configured to prevent formatting errors or abnormal value ranges.

[0026] This step allows the system to obtain an individualized dataset covering all the key variables required for drug modeling, providing an accurate, complete, and standardized data foundation for the subsequent construction and precise simulation of physiological pharmacokinetic models.

[0027] Step S102: Based on the individualized parameter dataset, construct a physiological pharmacokinetic model suitable for the target receptor, and set the physicochemical properties, tissue distribution parameters, metabolism and excretion parameters of the drug related to oxytocin injection to simulate the in vivo concentration change process under different administration conditions.

[0028] In step S102, the system constructs a physiological pharmacokinetic (PBPK) model specifically suited for the target receptor based on the individualized parameter dataset formed in step S101. This model simulates the dynamic process of oxytocin injection in the human body under different administration conditions, achieving a full simulation of the drug's concentration distribution, metabolism, and excretion in various tissues and organs.

[0029] The model employs a multi-compartment structure, dividing the human body into several functional compartments according to anatomical and physiological logic. Common compartments include plasma, liver, kidneys, uterine tissue, placenta, muscle, fat, and brain tissue. These compartments are connected by blood flow, forming a complete systemic transport pathway. Each compartment is modeled using a set of differential equations based on the law of conservation of drug mass, expressing the change in drug concentration within that compartment over time.

[0030] Under normal circumstances, the drug concentration C in a certain tissue compartment is... t The change in (t) can be expressed by the following formula:

[0031]

[0032] Among them, C t (t) represents the drug concentration in the tissue compartment at time t, in mg / L. This value is the objective of this equation; Qt represents the blood flow velocity in the tissue, in L / h. Its value can be obtained from a standard physiological parameter database and linearly scaled according to the target receptor's body weight, such as:

[0033]

[0034] Among them WT actual For the target receptor body weight, WT ref For reference to an individual's weight, Q t,ref For reference individuals, the tissue blood flow rate is used.

[0035] V t This indicates the volume of the tissue, expressed in liters (L). Similar to Q. t Weight can also be adjusted after obtaining it from a standard database;

[0036] C b (t) represents the drug concentration in the blood at time t, in mg / L. This value is derived from the calculation output of the plasma compartment.

[0037] K p The tissue-blood partition coefficient is a dimensionless coefficient. This value can be calculated using the following empirical formula:

[0038]

[0039] Among them, f IW The proportion of intracellular water in the tissue; f EW The proportion of extracellular water in tissues; f NL The proportion of neutral lipids in the tissue; P owThe octanol-water partition coefficient of the drug; BP is the plasma protein binding ratio of the drug; these parameters are derived from pharmacopoeia databases, literature values, or experimental determinations;

[0040] CL int，t This indicates the intrinsic clearance rate of the drug by the tissue, expressed in L / h. It is non-zero for metabolically relevant organs (such as the liver and kidneys), and can be set to 0 for other tissues. Its value needs to be estimated based on individual liver enzyme activity (such as ALT and AST) or kidney function indicators (such as eGFR). For example, for liver metabolism, it can be expressed as:

[0041]

[0042] Among them CL ref ALT is the standard hepatic clearance rate. actual The value of alanine aminotransferase (ALT) is the target receptor value, α is the empirical decay factor (usually set to 0.5–1), and ALT is the ALT value. ref The alanine aminotransferase level is used as a reference for individuals.

[0043] For renal excretion, the clearance rate is calculated using the glomerular filtration rate (GFR), commonly using the following formula:

[0044] CL renal =f u ·GFR

[0045] Among them, f u The free percentage of the drug (1 - plasma protein binding rate) has no unit; GFR is the glomerular filtration rate, in L / h.

[0046] In the model, the plasma compartment, tissue compartment, and uterine tissue compartment are connected by blood flow to form a systematic network. The master solver solves the aforementioned differential equations simultaneously using time recursion. For the modeling of uterine tissue, a simplified pharmacodynamic module can be embedded in its compartments to predict the concentration-effect relationship. This part can use the Sigmoid Emax model.

[0047]

[0048] Where E(t) represents the predicted value of uterine contraction frequency or intensity; E max This represents the maximum uterine contraction response, measured in contractions per minute, and can be set based on clinical data or obstetric literature.

[0049] EC 50 The half-maximum effect concentration is expressed in mg / L; n is the Hill coefficient, describing the steepness of the response curve. It is usually set to 1–3.

[0050] C t (t) represents the drug concentration in uterine tissue, calculated using the PBPK model.

[0051] The PBPK model can be constructed using platforms such as Matlab SimBiology, PK-Sim, Simcyp, or custom Python scripts. The specific implementation should include a model structure module, a parameter assignment module, a numerical solution module, and a result output module. To accommodate the differences among target individuals, all parameter assignment processes must receive the individualized parameter dataset obtained in step S101 as input, and perform logical mapping and dynamic replacement to ensure that each model instance uniquely corresponds to the target receptor.

[0052] Ultimately, the model will output the concentration curves of oxytocin in each tissue compartment over time under given dosing parameters (such as injection rate and time), with particular attention to the local peak concentration in the uterine tissue and the systemic plasma concentration, providing a reliable basis for subsequent regimen screening.

[0053] In summary, this step clearly completed the construction of the PBPK model through a well-structured modeling framework, precise parameter configuration, and mathematical solution path, ensuring that the model possesses sufficient individual adaptability. For clinical validation, this model can be further compared with the recipient's actual blood drug concentration or uterine contraction frequency monitoring data to verify its accuracy and robustness.

[0054] Furthermore, based on the individualized parameter dataset, a physiological pharmacokinetic model suitable for the target receptor is constructed, and the physicochemical properties, tissue distribution parameters, metabolic and excretion parameters related to oxytocin injection are set to simulate the in vivo concentration changes under different administration conditions, including:

[0055] Based on the individualized parameter dataset, perform individual parameter standardization processing, including unifying units, normalizing reference values, and imputing missing values ​​for weight, height, liver and kidney function indicators, and placental blood flow parameters, to generate a standardized parameter input vector that fits the modeling format.

[0056] Using the standardized parameter input vector, a structured physiological model of multiple organ compartments is constructed. The model includes at least six functional compartments, including the uterus, liver, kidneys, systemic circulation, placenta and fetus. Each compartment is represented by a system of differential equations to represent the material migration process, and cross-organ connection paths and blood flow rate weights are established.

[0057] Based on the organ partitioning model, the drug physicochemical property library of oxytocin injection is called to automatically match the drug partition coefficient, tissue affinity coefficient and metabolic enzyme action pathway, and the metabolic clearance rate and renal excretion constant are calculated in combination with individualized liver and kidney function levels to form a parameterized PBPK model for this target receptor.

[0058] Based on the parameterized PBPK model, a dynamic physiological simulation scenario is constructed. Multiple sets of simulation conditions are set, including different intravenous injection rates, variable speed strategies, and cyclic stress states. Time-step simulation is executed, and drug concentration distribution data in each organ compartment at each time point are output for subsequent concentration-response relationship evaluation and dosing parameter screening.

[0059] In this invention, in order to construct a physiological pharmacokinetic model (PBPK model) suitable for the target receptor based on an individualized parameter dataset of the target receptor, and to accurately set the physicochemical properties, tissue distribution parameters, metabolic and excretion parameters of the drug related to oxytocin injection, so as to simulate the dynamic concentration change process of oxytocin in vivo under different administration conditions, it is necessary to carry out modeling and simulation environment setting in stages to ensure that the obtained model has sufficient individual adaptability and spatiotemporal resolution, thereby providing a reliable concentration prediction basis for precise drug administration.

[0060] First, the collected individualized parameter dataset typically includes raw clinical data such as weight, height, age, liver function indicators (e.g., ALT, AST, liver enzyme activity), kidney function indicators (e.g., creatinine clearance, blood urea nitrogen level), placental blood flow, and reproductive system status (e.g., gestational age, cervical maturity). To ensure that this data can be uniformly used in the PBPK modeling process, this embodiment of the invention first performs standardization processing. This processing includes three sub-steps: First, unit standardization, which converts all parameter values ​​to international standard units that meet the calculation requirements of the PBPK model (e.g., weight is standardized to kilograms, blood flow to ml per minute); second, reference value normalization, which divides the parameter values ​​by the population mean reference value to obtain dimensionless standardized coefficients, thereby eliminating model mismatch problems caused by different units and magnitudes; third, missing value imputation, which, when some clinical data is missing, uses the population mean, correlations between other parameters of the subjects, and machine learning regression algorithms to imputate and fill in missing data, ensuring the generation of a complete standardized parameter vector.

[0061] Based on the standardized individual input vectors described above, the model enters the structured physiological modeling stage involving multiple organ compartments. In this stage, the PBPK model is designed to consist of at least six organ or tissue functional compartments, including the uterus, liver, kidneys, circulatory system, placenta, and fetus. Each compartment is modeled as a simulated structure corresponding to the anatomy and physiological properties of the actual organ. Within this structure, drug concentration changes in the tissue are expressed using a differential approach. Inter-compartmental material transport processes are constructed by analogy to blood perfusion, including the first-pass pathway from blood to the liver, the delivery pathway from circulatory system to the uterus, and metabolic and excretion pathways in the liver and kidneys. Furthermore, each pathway is weighted according to the proportion of blood perfusion, and differences in blood flow distribution among individuals are directly reflected in the connectivity parameters, ensuring individualized model representation.

[0062] After defining the organ structure and connectivity network, each tissue compartment needs to be further assigned pharmacokinetic properties related to oxytocin injection. This invention provides a structured drug property library, containing multiple key parameters such as the molecular weight of oxytocin, lipid-water partition coefficient, plasma protein binding rate, hydrolytic enzymatic rate, and receptor affinity constant. The system automatically retrieves the parameters required for drug distribution in each tissue based on the organ structure of the target model, such as the tissue affinity coefficient of the uterus, the placental transport rate coefficient, and the liver's metabolic capacity parameters, and dynamically adjusts the metabolic clearance and excretion rates based on the individual's liver and kidney function levels. For example, in individuals with renal insufficiency, the excretion constant will be correspondingly reduced; while in individuals with elevated liver enzyme activity, the liver metabolic rate will be increased. All pharmacokinetic parameters are automatically embedded into the dynamic migration module of each compartment in a structured configuration, thereby constructing a parameterized PBPK model for a specific receptor individual.

[0063] After initializing the model parameters, the simulation phase begins. This invention constructs a multi-scenario physiological simulation environment to simulate drug distribution under different dosing strategies. Specifically, the system supports setting multiple sets of intravenous injection rates, including switching modes between initial and constant rates, and allows the introduction of variable-rate strategies during the simulation, such as dose adjustment when uterine sensitivity changes. Furthermore, to improve the physiological realism of the model, it also allows setting stress state parameters of the circulatory system, including changes in blood pressure, heart rate, and end-perfusion, based on which a multi-dimensional simulation scenario is constructed.

[0064] The simulation employs a time-stepping approach, whereby at each simulation moment, the drug concentration values ​​in each compartment are updated sequentially, and mass balance calculations are performed based on the current dosing rate and inter-compartment migration rate. The model records the trajectory of drug concentration changes in all organs from the start time to the target time period, forming a data matrix containing multiple dimensions such as uterine concentration, systemic circulating concentration, and liver concentration. This matrix serves as the input source for subsequent pharmacokinetic response evaluation and dosage optimization, providing high-precision, high-temporal-resolution pharmacokinetic support.

[0065] In summary, the PBPK modeling and simulation pathway provided by this invention not only realizes a fully traceable modeling mechanism from individual physiological parameters to simulation output, but also enables the model to flexibly adapt to changes in drug administration conditions for different individuals and pathways through modular parameter settings and multi-compartment structural decomposition, ensuring accurate reflection of the distribution and dynamic concentration of oxytocin injection in vivo, and providing a solid foundation for concentration-response relationship modeling and drug administration parameter screening in subsequent steps.

[0066] Step S103: Input multiple sets of drug administration parameters, including injection initiation rate, maintenance rate and duration of administration, into the physiological pharmacokinetic model, perform simulation calculations to obtain the corresponding concentration change curves, and evaluate the drug concentration level, duration of administration and predicted uterine contraction response of each set of drug administration parameters in uterine tissue and systemic circulation.

[0067] In step S103, the system, based on the pre-constructed and parameterized physiological pharmacokinetic (PBPK) model, inputs multiple sets of set dosing parameters to conduct simulations of the intravenous behavior of oxytocin injection, and analyzes the simulation results to evaluate the distribution, efficacy, and safety of different regimens in the target receptor. This step is not only a major application step after model establishment, but also the core foundation for subsequent screening of optimal dosing parameters.

[0068] First, at the input level, a candidate set of multiple dosing parameters needs to be constructed. These "dosing parameters" refer to the control variables of injection behavior in the time and dose dimensions, typically including the injection initiation rate, variable rate strategy, maintenance rate, and total dosing duration. To meet the model's analytical requirements for continuous-time input, the injection rate should be constructed as a function of time, such as a linearly increasing, stepwise increasing, constant rate, or a more complex piecewise function. Each set of dosing parameters will represent an independent simulation path, and their effectiveness will be tested one by one in the PBPK model.

[0069] At the execution level, each set of dosing parameters is input as an external signal to the "drug administration event" interface in the PBPK model. This interface controls the rate function of drug entry into the plasma compartment. The model calculates the amount of drug per unit time based on the current input rate at each moment and incorporates this amount into the mass conservation differential equation for plasma compartment concentration. The system uses an ODE (Ordinary Differential Equation) solver as its core computational engine, typically employing the Runge-Kutta method, LSODA method, or other variable-step-size adaptive algorithms to ensure high-precision tracking of the dynamic evolution of drug concentration across the entire time domain. The model's time step should be sufficiently small to reflect instantaneous changes in the injection rate, while the total simulation duration should cover the entire injection process and part of the elimination period to reflect the complete changes in peak and maintenance concentrations.

[0070] During the simulation, the PBPK model will output time-series concentration change curves of oxytocin in each compartment, especially the data from the uterine tissue compartment and the plasma compartment, which will be the focus of this step of analysis. Based on the concentration change curves, the system needs to further calculate three core response indicators: peak concentration and maintenance concentration in the uterine tissue, maximum plasma concentration in the systemic circulation, and the duration of achieving an effective concentration. The peak concentration in the uterine tissue will be used to measure whether the regimen has reached the set onset threshold in terms of efficacy, the upper limit of the systemic circulation concentration will be used to determine whether there is a potential risk of overdose, and the duration can be used to reflect whether the dosing regimen has established a stable effect within the clinically permissible intervention time.

[0071] In addition to pharmacokinetic parameters, the model also needs to assess the differences in pharmacodynamic response among each regimen. For this purpose, a pharmacodynamic (PD) response module is typically embedded in the uterine tissue compartment of the PBPK model. This module receives the time series of drug concentrations in that compartment and outputs predicted uterine contraction frequencies using a Sigmoid Emax model. Its mathematical expression is:

[0072]

[0073] Where E(t) represents the predicted frequency of uterine contractions at a given time point (e.g., the number of contractions per 10 minutes), C t (t) represents the current concentration in uterine tissue, E max For the maximum achievable contraction frequency, EC 50 The concentration required to produce half of the maximum effect, where n is the Hill coefficient, which controls the slope of the reaction curve.

[0074] The predicted uterine contraction frequency will form a complete pharmacodynamic response curve, from which the system can further extract key pharmacodynamic parameters, such as the shortest time required to reach the target uterine contraction frequency and the length of time during which the uterine contraction frequency exceeds the safety threshold. These derived indicators are used to quantify the rate, intensity, and stability of the pharmacodynamic response.

[0075] All simulation data will be output in a structured data format (e.g., CSV or database table) after each parameter test, storing the corresponding input parameter group number, pharmacokinetic indicators, efficacy prediction results, etc. The system can be set up with an automatic labeling mechanism to mark parameter groups that meet the initial screening criteria as "candidate solutions" to facilitate subsequent steps such as cluster analysis or multi-objective decision optimization.

[0076] It is important to note that the simulated dosing parameter set should cover a possible range of real-world clinical scenarios. For example, the injection rate should be set between 0.5 mU / min and 20 mU / min, the total dosing time can be set between 30 minutes and 3 hours, and the variable-rate regimen should at least cover three typical clinical pathways: constant rate, gradual escalation, and on-demand adjustment. The parameter set can be generated using Latin hypercube sampling (LHS) or a full factorial combination method to ensure the representativeness of the parameter space.

[0077] In summary, step S103 is not only a static run of the PBPK model, but also a dynamic simulation and response analysis process based on multiple input variables. This process, through high-frequency, grouped simulation tests, achieves a quantitative evaluation of different dosing regimens in both pharmacokinetic and pharmacodynamic dimensions, providing a clear, specific, and systematic basis for subsequent regimen selection.

[0078] Furthermore, the process involves inputting multiple sets of dosing parameters, including injection initiation rate, maintenance rate, and dosing duration, into the physiological pharmacokinetic model, performing simulation calculations to obtain corresponding concentration change curves, and evaluating the drug concentration levels, dosing duration, and predicted uterine contraction responses for each set of dosing parameters in uterine tissue and systemic circulation, including:

[0079] An initial set of drug administration parameters is constructed based on the injection initiation rate, maintenance rate, and drug administration duration. Then, using a perturbation screening mechanism based on drug sensitivity weights, combinations of injection initiation rate, maintenance rate, and drug administration duration with high response gradients are extracted from the initial set of drug administration parameters to form a candidate set of drug administration parameters.

[0080] The candidate combination of dosing parameters is input into the physiological pharmacokinetic model item by item, and individualized weight perturbation values ​​of placental blood flow parameters and liver and kidney function indicators are introduced in each simulation operation to simulate the effect of blood flow redistribution of receptors on drug concentration time sequence under real physiological variations of pregnancy, and obtain the set of concentration change curves of each combination of dosing parameters in uterine tissue and system circulation.

[0081] Based on the set of concentration change curves and the concentration-effect relationship on which the predicted uterine contraction response depends, the predicted uterine contraction response for each group is calculated using a multi-parameter dynamic function with a time delay factor, based on the injection initiation rate, maintenance rate and administration duration. A three-dimensional pharmacodynamic index matrix including the peak concentration of uterine tissue, the systemic circulating exposure area and the response delay duration is output.

[0082] A dual-objective clustering process is performed on the three-dimensional pharmacodynamic index matrix. The advantageous solution set is extracted according to the principle of maximizing the concentration level in uterine tissue and minimizing the drug concentration level in the system circulation. Combined with the drug administration duration as a constraint, a structured target drug administration parameter mapping table is formed.

[0083] Based on the target dosing parameter mapping table, a cross-individual variable perturbation test scheme was designed. Multiple sets of simulated perturbations, including the range of changes in body weight, placental blood flow parameters, and liver and kidney function indicators, were input into the physiological pharmacokinetic model. Simulation stability verification was performed for each set of target dosing parameter combinations. The combination of injection initiation rate, maintenance rate, and dosing duration that can maintain the drug concentration level in the uterine tissue at no lower than the effective concentration, the drug concentration level in the system circulation at no more than the safety limit, and the predicted value of uterine contraction response within the target range under all perturbation conditions was retained as the final recommended dosing parameter results.

[0084] In implementing the oxytocin injection precision administration method based on the PBPK model optimization described in this invention, to achieve dynamic response regulation of individual receptors, the system needs to model, simulate, and evaluate multiple potential dosing regimens to select the optimal parameter combination that meets clinical efficacy and safety boundaries. This description focuses on the specific implementation path of "inputting multiple sets of dosing parameters, including injection initiation rate, maintenance rate, and dosing duration, performing simulation calculations to obtain corresponding concentration change curves, and evaluating the drug concentration level, dosing duration, and predicted uterine contraction response of each set of dosing parameters in uterine tissue and systemic circulation," especially how to construct dosing parameter combinations, how to introduce individualized variation factors for simulation, and how to extract the final recommended parameters under the highly complex physiological conditions of human pregnancy.

[0085] First, the system needs to construct an initial set of dosing parameters for exploratory simulation within a preset parameter range. This set consists of three dimensions: injection initiation rate, maintenance rate, and dosing duration. During construction, the injection initiation rate is set to cover multiple gradients within a clinically common range (e.g., 0.5-10 mU / min), the maintenance rate is set to several stable segments (e.g., 2-20 mU / min), and the dosing duration is set to different duration intervals (e.g., 30 minutes to 6 hours). The system uses parameter space coverage strategies such as grid sampling or Latin hypercube sampling to ensure that the set is evenly distributed and complete in the multidimensional variable space, thus facilitating subsequent response screening and optimization.

[0086] To avoid redundant calculations and improve screening efficiency, the system implements a perturbation screening mechanism based on drug sensitivity weights on the initial set of dosing parameter combinations. This mechanism, leveraging prior clinical data or simulation data, establishes a scoring function reflecting the sensitivity of oxytocin's response to uterine tissue concentrations at different dosing rates. The scoring function can be comprehensively evaluated by calculating indicators such as the maximum rate of concentration change, the initial rise gradient, and the response slope within the lag time window during the simulation. Combinations with higher scores are more sensitive to dosing rates and are more likely to produce clinically significant effects, thus being prioritized for inclusion in subsequent simulation calculations. The system ultimately extracts combinations of injection initiation rate, maintenance rate, and dosing duration with high response gradients from the initial set of combinations, forming a concise but more representative and discerning set of candidate dosing parameter combinations.

[0087] Next, the system inputs each parameter from the candidate combination set of dosing parameters into the physiological pharmacokinetic model. The physiological pharmacokinetic model consists of multiple functional compartments, covering key structures such as uterine tissue, liver, kidneys, placenta, systemic circulation, and fetus. During the simulation, to reflect pharmacokinetic fluctuations due to individual differences, the system introduces individualized weighted perturbation values ​​for placental blood flow parameters and liver and kidney function indicators in each simulation. Specifically, the system sets a perturbation vector representing individual physiological variations. This vector covers key indicators such as placental perfusion rate, uteroplacental blood flow distribution ratio, liver enzyme metabolic efficiency, and glomerular filtration rate. Linear, exponential, or Bayesian-based uncertainty perturbations are added to the original values ​​to make each simulation more closely resemble the actual individual differences in pregnancy. The perturbation range is typically set with upper and lower limits (e.g., ±20%) based on clinical data to cover common scales of variation in the actual population. Through this mechanism, the concentration change curve obtained by the system is no longer a static output under ideal conditions, but a set of concentration fluctuation data with realistic dynamic tolerance.

[0088] Each set of drug administration parameters was subjected to a complete simulation under the aforementioned perturbation. The simulation period was typically 0 to 6 hours, with time steps ranging from 30 seconds to 1 minute. During the simulation, the system iteratively calculated the drug concentration values ​​for each organ compartment in a time-series manner, paying particular attention to the time-series concentration change curves in the two key compartments of uterine tissue and systemic circulation. All simulation results were collected into a structured dataset, where each set of parameters corresponds to a set of concentration curves, labeled in time series format, forming a set of concentration change curves.

[0089] After obtaining the set of concentration change curves, the system further combines the concentration-effect relationship upon which the predicted uterine contraction response depends to construct a multi-parameter dynamic function with a time delay factor. This dynamic function is used to calculate the predicted uterine contraction response value corresponding to each group of injection initiation rate, maintenance rate, and administration duration. In this function, the system first sets an Emax or Sigmoid Emax function based on pharmacodynamic principles, where drug concentration is used as the input variable, and uterine contraction frequency, intensity, or uterine contraction response probability is used as the output result. To adapt to the time lag characteristics of oxytocin, the system introduces a delay factor inside the function, such as the Lag-time or Transit compartment mechanism, to simulate the delayed links in the drug action mechanism, such as the activation process of uterine smooth muscle receptors and the time consumed by signal transduction. During the simulation, this function combines with concentration data to dynamically calculate the estimated value of the uterine contraction response within each time step, and finally integrates the response indicators of a single set of parameters in three dimensions: peak uterine tissue concentration, system circulating exposure area (AUC), and response delay duration, forming a three-dimensional pharmacodynamic index matrix.

[0090] To extract the optimal combination of dosing parameters, the system performs bi-objective clustering on the three-dimensional pharmacodynamic index matrix. The core principle of this clustering is to maximize the concentration level in uterine tissue while minimizing drug concentration exposure in the system circulation, thus seeking a balance between efficacy and safety. The clustering algorithm can employ a Pareto front-based or K-means-based objective ranking mechanism to automatically identify a series of parameter combinations that perform optimally in three-dimensional space, forming a set of advantageous solutions. The system further incorporates the dosing duration as a constraint, excluding combinations that do not meet the dosing time limit setting, and finally outputs a structured target dosing parameter mapping table. This mapping table clearly indicates the injection initiation rate, maintenance rate, and dosing duration values ​​for each advantageous combination, along with their corresponding pharmacodynamic evaluation indicators, and serves as a candidate set for subsequent validation and recommendation.

[0091] However, to ensure the stability and robustness of the combinations in the target dosing parameter mapping table under real clinical conditions, the system also needs to execute a cross-individual variable perturbation test protocol. This test protocol designs multiple perturbation simulation conditions, including but not limited to the range of weight variation during different stages of pregnancy, the high and low limits of placental blood flow parameters, simulated values ​​of liver enzyme activity changes with pregnancy, and the range of glomerular filtration rate variations. The system combines these variables into a set of perturbation inputs and sequentially loads these inputs into the physiological pharmacokinetic model, performing cross-simulations with each combination in the target dosing parameter mapping table to form a complete stability verification process.

[0092] In each simulation round, the system determines whether the combination can maintain the efficacy target under the perturbation state, including: the drug concentration level in uterine tissue is not lower than the set effective concentration, the drug concentration level in the system cycle does not exceed the clinical safety limit, and the predicted uterine contraction response remains within the treatment target range. Only combinations that meet the above three conditions under all perturbation scenarios will be marked by the system as acceptable final recommended dosing parameters. If a combination fails under some perturbation conditions, it will be removed or re-enter the adjustment cycle.

[0093] The final system outputs recommended dosing parameters, including several combinations of injection initiation rate, maintenance rate, and dosing duration validated through perturbation testing. Each combination is accompanied by detailed concentration curves, response indicators, and applicable perturbation ranges, allowing clinicians to match and select the appropriate parameters based on the patient's physiological information, ensuring a precise, safe, and efficient oxytocin injection dosing regimen.

[0094] Step S104: Select the target dosing parameter combination from the simulation results that simultaneously meets the following clinical control conditions: the drug concentration in the uterine tissue reaches the set effective concentration, the drug concentration in the system circulation does not exceed the set safety limit, the dosing duration is within the set range, and the predicted value of uterine contraction response is within the target range.

[0095] In step S104, the system's task is to comprehensively screen and evaluate all simulation results obtained through PBPK model simulation, and select one or more sets of target dosing parameter combinations that simultaneously meet the preset clinical control requirements. This step is a key link in the model-driven dosing optimization process, and its goal is to establish a screening mechanism that is highly consistent with clinical efficacy and safety from both pharmacokinetic (PK) and pharmacodynamic (PD) dimensions.

[0096] First, after the PBPK model completes simulations of multiple combinations of drug administration parameters, the system has obtained a large amount of data in time-series format. The core data includes oxytocin concentration change curves in uterine tissue, systemic plasma concentration change curves, and uterine contraction frequency prediction curves driven by uterine concentration inputs. This data is stored in structured arrays or databases, with each record corresponding to a set of input parameters, including injection initiation rate, maintenance rate, and total drug administration duration.

[0097] The system then needs to calculate three key evaluation indicators for each simulation result: 1. Whether the drug concentration in the uterine tissue reaches the clinically required effective concentration; 2. Whether the drug concentration in the system plasma is within the safety limit; 3. Whether the uterine contraction response is within the expected target frequency range throughout the entire dosing cycle. To achieve this screening, the upper and lower limits of each screening criterion must be defined at the initial stage of model building, and all screening should be performed in numerical form to avoid ambiguous logical judgments.

[0098] For the first screening criterion, namely, the drug concentration in uterine tissue reaching the effective concentration, the system needs to analyze the concentration curve C of that tissue compartment. uterus , calculate its maximum value C over the entire simulation time interval. uterus,max This value should be greater than or equal to the set effective concentration threshold C. threshold，efficacy Otherwise, the regimen is considered unable to establish an effective uterine response in terms of efficacy. For example, if the effective concentration set by the literature or clinical expert system is 0.02 ng / mL, all regimens that do not reach this value in the simulation are eliminated.

[0099] For the second screening indicator, namely whether the drug concentration in the system plasma exceeds the standard, it is necessary to analyze the concentration curve C in the plasma compartment. plasma (t), extract its maximum value C plasma，max And compare it with the set system security threshold C threshold，safety Comparison. This safety threshold is determined based on previous clinical data or adverse drug reaction databases. For example, the risk threshold for events such as hypotension and fetal distress caused by oxytocin is usually set at no more than 0.15 ng / mL in the literature. Simulation protocols exceeding this value are considered to have a systemic toxicity risk and should be excluded.

[0100] The third screening criterion involves controlling the frequency of uterine contractions. The system needs to analyze the uterine contraction frequency prediction curve F(t) output by the Emax pharmacodynamic model. This analysis requires that the predicted frequency not only reach the target range [F] within a certain time period. target -ΔF, F target [+ΔF] also requires that the frequency be maintained within the target range for a duration that is not too short, to ensure that contractions are both strong and sustained. At this point, the system typically sets a minimum maintenance time T. min For example, the duration should be no less than 15 minutes, and a sliding window method should be used to count the total duration of the frequency within the target range over a continuous period. Only plans that meet both conditions of "entering the target frequency range" and "sufficient maintenance time" will be considered qualified in terms of uterine contraction response.

[0101] The above three indicators are hard criteria. The system can construct logical judgment expressions to batch mark qualified and unqualified items in the simulation result database. Qualified solutions will be retained, along with the original input parameters, drug concentration curves, and predicted uterine contraction response values ​​as subsequent output records. If multiple solutions are satisfied simultaneously, the system can also set optimization and ranking rules, such as prioritizing those with shorter dosing times, smaller drug doses, or the lowest system concentration peaks, to further filter out an optimal solution.

[0102] In this step, all judgment criteria should be explicitly quantified and expressed without relying on subjective human judgment to ensure the reproducibility and consistency of the screening mechanism. Those skilled in the art can use structured data analysis tools (such as SQL databases, Python scripts, or MATLAB programs) based on the above indicator framework to automate the processing of large-scale simulation results, thereby obtaining accurate, reliable, and compliant target dosing regimens. These regimens will directly proceed to the next step as the basic input for generating injection control instructions.

[0103] In summary, step S104 achieves the entire process of selecting the optimal dosing regimen from large-scale simulation results by establishing three non-redundant clinical control indicators that are highly correlated with efficacy and safety.

[0104] Furthermore, the selection of target dosing parameter combinations from the simulation results that simultaneously meet the following clinical control conditions—drug concentration in uterine tissue reaching a set effective concentration, drug concentration in systemic circulation not exceeding a set safety limit, dosing duration within a set range, and predicted uterine contraction response within a target interval—includes:

[0105] Extract the numerical sequence of drug concentration changes over time in uterine tissue for each combination of injection initiation rate, maintenance rate and administration duration from the simulation results, and identify the time point at which the set effective concentration is first reached, as the reference time point when the combination begins to exert physiological effects in the target receptor.

[0106] Based on the reference time point, extract the drug concentration change data in the corresponding system cycle, calculate the cumulative concentration exposure value of the combination within a set time window after the reference time point, and compare it with the set safety limit. Exclude all combinations of injection initiation rate, maintenance rate and administration duration whose cumulative concentration exposure value exceeds the set safety limit.

[0107] For combinations of injection initiation rate, maintenance rate and administration duration that were not excluded, the curves of the corresponding predicted uterine contraction response over time were further analyzed. Multiple evaluation indicators, including response onset time, peak response amplitude, peak duration and response stability, were extracted to quantitatively assess the physiological response performance of all combinations.

[0108] Based on the results of quantitative assessment, combined with the judgment results of whether the drug concentration in uterine tissue reaches the set effective concentration, whether the drug concentration in the system circulation does not exceed the set safety limit, whether the duration of administration is within the set range, and whether the predicted value of uterine contraction response is within the target range, the combination of injection initiation rate, maintenance rate and duration of administration that meets all clinical control conditions is selected as the target administration parameter combination.

[0109] In the precise dosing method for oxytocin injection optimized based on the PBPK model, a key step is to screen out the target dosing parameter combination that simultaneously meets multiple clinical control conditions from the simulation results. This process involves not only analyzing the concentration change patterns of the drug in the target tissue and systemic circulation, but also fully considering the predictive efficacy of the injection strategy on uterine contraction response, and forming quantifiable and screenable judgment criteria under multiple evaluation dimensions to guide the final determination of dosing parameters.

[0110] First, for each combination of injection parameters obtained in the simulation, a data sequence of drug concentration changes over time within the uterine compartment must be extracted. This data sequence is generated by the PBPK model under set simulation conditions, based on individualized physiological parameters and pharmacokinetic characteristics. The concentration change sequence is generally recorded at a fixed time step (e.g., once per minute) until the simulation ends. Within this concentration sequence, the system compares each sequence point by point by setting a predefined threshold for the effective concentration. Once the concentration at a certain time point exceeds the threshold for the first time, that time point is recorded as the "physiological onset time" of the dosing parameter combination in the receptor. This reference time point can be used for subsequent assessment of the onset time of the effect, thus serving as an important indicator for determining whether the drug takes effect within a reasonable timeframe.

[0111] Secondly, based on the aforementioned physiological reference start time, the system further extracts the corresponding systemic circulation zone's drug concentration change data. By accessing the data recording channel of the corresponding systemic circulation zone in the PBPK model, it retrieves the concentration data sequence within a set time period (e.g., 30 minutes, 60 minutes) after that start time point, and calculates the cumulative concentration exposure value based on this sequence. This value reflects the overall level of systemic drug exposure and is usually calculated through integral calculation or weighted averaging. To ensure that the drug concentration in the systemic circulation does not exceed the set toxicity risk range, the system compares this cumulative exposure value with the set safety limit. All dosing parameter combinations exceeding this limit are marked as "excluded" and no longer participate in subsequent uterine contraction response assessments.

[0112] Subsequently, for any combinations of dosing parameters that were not excluded, the system included their corresponding uterine contraction response prediction curves in the analysis. The predicted uterine contraction response value is derived from the concentration-response mapping relationship established historically in PBPK simulations of drug concentration in uterine tissue, and generates a dynamic response curve over time. During the actual analysis, the system extracts several key indicators from this response curve, including: onset time (i.e., the time required for the predicted response value to first exceed the onset response threshold), peak response amplitude (the maximum response value reached by the predicted value), peak duration (the duration of the maximum response value), and response stability (the degree of fluctuation of the response curve within the effective range). These indicators are derived by finding time points in the curve, calculating amplitude differences, and analyzing the rate of change of the response trend, forming a quantifiable set of features.

[0113] Based on the extracted set of pharmacodynamic indicators, the system maps the performance of each combination to a multidimensional evaluation space and filters them one by one according to the set control conditions. First, it determines whether the drug concentration in the uterine tissue meets the set effective concentration condition; second, it determines whether the circulating drug concentration in the system is within the set safety limits; third, it determines whether the dosing duration is within the acceptable dosing duration range for the target receptor (e.g., not exceeding 8 hours or other clinically defined parameters); finally, it determines whether the predicted uterine contraction response value stably falls within the preset target range (e.g., the target range can be set based on clinical studies, for example, the predicted response value is between 0.6 and 0.85). Only when a combination meets the conditions in all four dimensions is it marked by the system as a "clinically acceptable target dosing parameter combination".

[0114] It should be noted that, in the actual implementation process, to improve evaluation efficiency, the system typically incorporates a batch processing mechanism to process a large number of simulation results at once, and achieves automatic filtering through parameterized conditional screening templates. These screening templates not only include numerical judgment criteria such as concentration thresholds and safety limits, but also include trend analysis algorithms for the response curve structure, such as identifying inflection points and determining whether the response fluctuation range exceeds stability judgment criteria. These algorithms can be pre-trained using known physiological feature samples or designed through rule extraction methods. The key is to ensure accurate differentiation between "effective response" and undesirable combinations of "large fluctuations and uncontrollable efficacy."

[0115] Furthermore, to ensure the clinical stability and adaptability of the final screening results, the system also needs to perform cross-individual stability validation on the target dosing parameter combinations. This process involves inputting perturbations of individual variables such as body weight, placental blood flow parameters, and liver and kidney function indicators into the original PBPK model, and re-running the simulation for each target combination under multiple perturbation conditions. If a combination remains effective under different individual variable perturbations within the above four control conditions, then the combination will be marked as a "strongly robust target dosing parameter combination" and can be used as the final recommended dosing regimen.

[0116] Through the above operational path, this step not only systematically integrates simulation output, clinical condition judgment, and response curve analysis, but also achieves the quantitative extraction and screening of complex pharmacodynamic response behaviors through structured data processing.

[0117] Step S105: Based on the selected target drug delivery parameter combination, generate injection control instructions, which include injection initiation rate, speed adjustment information, maintenance rate and termination time, and convert them into a format that can be recognized by the program-controlled injection device.

[0118] In step S105, the system needs to generate a set of injection control instructions that can be directly executed by clinical injection devices, based on the target dosing parameter combinations that have been simulated and analyzed by the PBPK model and screened through clinical control indicators. The core task of this process is to format and output the dosing strategy determined in the previous step—namely, the injection initiation rate, variable rate adjustment scheme, maintenance rate, and termination time—in a clearly structured, device-readable, and programmable control instruction form, thereby achieving a closed-loop transformation from model deduction to device execution. This step not only requires rigorous logical expression and precise timing control, but also must consider the communication protocol, command format, execution granularity, and the ability to resolve temporal continuity of the injection device. Therefore, a balance must be struck between hardware compatibility, data integrity, and medical safety.

[0119] In practical implementation, the first step should be to structurally define the selected combination of target dosing parameters. This combination typically includes four types of variables: first, the injection initiation rate, representing the initial injection rate value when the system begins dosing, usually measured in mU / min (milliunits per minute); second, the rate adjustment information, which describes whether the injection rate needs to change dynamically throughout the dosing process, such as increasing linearly over time, adjusting non-linearly based on uterine contraction feedback, or remaining constant; third, the maintenance rate, i.e., how the system should maintain a stable injection rate to maintain the target uterine concentration and uterine contraction response after establishing initial efficacy; and fourth, the termination time, indicating when the injection should automatically end. All these parameters need to be uniformly encapsulated in a readable format for transmission to the injection device, and the format should be compatible with currently widely used programmable infusion pumps, including but not limited to infusion pumps, microinjectors, or embedded control devices with USB / serial / Bluetooth interfaces.

[0120] To complete this encapsulation process, the system needs to use a control instruction template to map the above four types of variables to instruction fields. Taking the common JSON format as an example, a complete injection control instruction can be constructed as follows:

[0121] {

[0122] "InjectionID": "OX001",

[0123] "StartRate": 2.0

[0124] "Profile": "linearramp",

[0125] "RampDuration": 15

[0126] "MaintainRate": 8.0

[0127] "TotalDuration": 60

[0128] "Unit": "mU / min"

[0129] }

[0130] In this example, "StartRate" represents the initial injection rate, "Profile" describes the rate change pattern (e.g., linear ramp, constant, segmented), "RampDuration" specifies the time (in minutes) required to ramp up from the initial rate to the maintenance rate, "MaintainRate" is the constant injection rate during the maintenance period, "TotalDuration" defines the total duration of the entire injection process, and "Unit" ensures that the device can recognize the dosage unit of the instruction. When generating instructions, the system can dynamically insert control segments based on the physiological response of the target receptor. For example, if fine-tuning of the maintenance rate or early termination is required during injection, conditional trigger statements can be added to the instructions, achieving highly flexible process control.

[0131] To ensure the injection device can accurately parse and execute the control command, the system needs to further write the structured control command into the injection device's control module via a communication interface. This process depends on the communication protocols supported by the device, such as USB serial communication, Bluetooth BLE commands, serial CAN bus, or RS485 interface. The command converter packages the JSON-formatted control command into a byte stream format recognizable by the device's underlying layer and sends a rate setting command in real time during each dosing cycle. Based on the received timestamps and rate values, the device updates the drive motor or propulsion device at preset intervals to achieve continuous output of the drug solution.

[0132] In terms of safety, the system needs to embed protection mechanisms in the instructions, such as a maximum injection rate limit, automatic pump shutdown settings for hardware timeouts, and interruption triggered by external physiological feedback signals (such as uterine contraction frequency), to prevent over-injection or missed injection due to data transmission errors or equipment malfunctions. All instructions should include timestamps and verification fields so that the system can perform real-time verification and recording of each control cycle during execution, ensuring the traceability and compliance of the entire drug administration process.

[0133] Furthermore, to meet the diverse needs of clinical applications, the instruction generation system should also support batch output and protocol archiving. For example, for dosing protocols for the same patient group, instruction templates can be exported in batches, and injection records can be archived through the database management module for later reuse, review, or reanalysis.

[0134] In summary, step S105 is not just a simple process of "generating control signals from calculation results", but a complete closed-loop system covering data structuring, protocol conversion, device communication, security verification and real-time control.

[0135] Furthermore, based on the selected target dosing parameter combination, an injection control command is generated. This injection control command includes the injection initiation rate, rate adjustment information, maintenance rate, and termination time, and is converted into a format recognizable by the program-controlled injection device, including:

[0136] Before generating the injection control command, based on the concentration change curves corresponding to each parameter in the target dosing parameter combination, the time point when the drug concentration in the uterine tissue caused by the injection initiation rate first reaches the set effective concentration is marked on the simulation time axis, and the concentration change rate and the trend of the predicted value of uterine contraction response from this time point to the termination time are recorded.

[0137] Based on the trend of concentration change rate and uterine contraction response prediction value, the time period when the fluctuation of concentration change rate and uterine contraction response prediction value is greater than a preset threshold is identified, and within the time period, corresponding variable speed adjustment information is generated according to the direction of concentration change, injection rate change history and predicted response risk level. The variable speed adjustment information includes the specified rate increase or decrease and its execution duration.

[0138] Based on the injection initiation rate, the maintenance rate, the variable speed adjustment information, and the termination time, a complete injection behavior parameter sequence is constructed, and the injection behavior parameter sequence is format-encoded. The injection rate and its change timing and amplitude at each time period are embedded into a unified control command template and converted into a digital control code stream with multi-segment rate adjustment capability.

[0139] The digital control code stream is subjected to integrity verification and injection device interface adaptation processing to ensure that after the injection control command is transmitted to the syringe system with program control function, the oxytocin injection operation is executed precisely according to the injection start rate, the speed adjustment information, the maintenance rate and the termination time.

[0140] In the precise administration method of oxytocin injection disclosed in this invention, one of the key steps is to generate injection control instructions to guide the injection device in precise drug delivery based on the selected target drug delivery parameter combination. The core of this step lies in converting the four types of control parameters—injection initiation rate, speed adjustment information, maintenance rate, and termination time—selected through PBPK simulation into complete control instructions in a programmable form with high time resolution. These instructions are then used by the injection device with programmable control functionality, enabling automated, personalized, and dynamically adjustable control of oxytocin injection drug delivery.

[0141] Before generating the injection control command, the system first acquires complete simulated output data corresponding to the target dosing parameter combination, particularly the curve data of drug concentration in uterine tissue changing over time, system circulating drug concentration data, and the predicted trend of uterine contraction response values ​​over time. This data is obtained by simulating a series of different dosing parameter combinations and evaluating their pharmacodynamic responses, representing the dynamic process of drug action under different dosing parameter controls in the current individual's physiological state.

[0142] In this step, the first step is to identify key events on the simulation timeline. The system iterates through the sequence of drug concentration changes in uterine tissue corresponding to the target parameter combination over time to find the time point when the concentration first reaches the set effective concentration. This point is usually considered the starting point where the drug effect begins to take effect and is therefore marked as the drug onset time point. Subsequently, using this time point as a baseline, the system continues to track the changing trends of drug concentration in uterine tissue and predicted uterine contraction response over subsequent time periods, forming a dynamic window for generating injection control commands.

[0143] Within this dynamic window, the system analyzes the rate of change in drug concentration—that is, the change in drug concentration per unit time—and the slope of change in the predicted uterine contraction response, thereby assessing the sensitivity and stability of the drug's efficacy response. Specifically, if the drug concentration or the predicted uterine contraction response exhibits drastic changes during administration, the system will identify the corresponding time period as a high-risk variability interval. To avoid overstimulating the uterus or exceeding drug concentration limits, an injection rate adjustment operation will be applied within this high-risk variability interval, i.e., generating "rate adjustment information."

[0144] The generation process of variable rate regulation information first considers the direction of drug concentration changes. For example, if the concentration rises rapidly within a short period and tends to exceed the upper limit, the system will instruct the injection device to reduce the rate; conversely, it will increase it. Secondly, the system incorporates variable rate history—the rate regulation patterns observed during the simulation phase when similar physiological fluctuations occurred—as a reference for current regulation decisions. Furthermore, the system considers the risk level of the predicted response, such as whether the predicted value of uterine contractions is close to the set upper limit, to determine whether protective deceleration is necessary. Finally, each high-risk time period is assigned a specific variable rate regulation instruction, which includes the direction of the rate change (increase or decrease), the magnitude of the change (e.g., increasing by 2 mL / h or decreasing by 1.5 mL / h), and the duration of the adjustment (e.g., 3 minutes, 5 minutes, etc.). This regulatory information constitutes a key component of the "variable rate regulation information" in the injection control instructions.

[0145] Next, the system integrates the injection initiation rate, maintenance rate, and termination time from the target dosing parameter combination with the previously generated multiple variable rate adjustment information to construct a complete parameter sequence for injection behavior. This parameter sequence does not simply contain four basic parameters, but rather a time-segmented rate control table, formatted as a set of "time point-rate value" pairs. Specifically, the system divides the entire dosing time into several control units (e.g., every 5 seconds or every 10 seconds), defining the injection rate value that should be maintained within each unit of time. For example, the initial 30 seconds are the injection initiation rate of 1.0 mL / h, the next 5 minutes are the maintenance rate of 1.2 mL / h, and then, due to fluctuations in the predicted value, a rate adjustment phase is entered, increasing the rate to 1.5 mL / h and maintaining it for 2 minutes, then decreasing it to 1.1 mL / h and maintaining it until the termination time. The generation of the entire parameter sequence is based on simulation results and risk control logic to ensure its clinical suitability and controllability.

[0146] The constructed sequence of injection behavior parameters is converted into a format recognizable by the device by the system. First, the sequence undergoes format encoding, whereby the rate value, start and end time, and control type (e.g., constant rate or adjustable rate) defined in each control unit are assembled into standardized data segments according to the command syntax required by the device. The system combines multiple segments into an injection control command stream, with each control command containing key parameters such as execution time, target rate value, and duration. Subsequently, the system packages this control command stream into a complete data structure and embeds it into a control command template, forming a digital control code stream with multi-segment rate adjustment capabilities. This code stream possesses robust time alignment and fault tolerance mechanisms, supporting safe execution in complex clinical scenarios.

[0147] After the code stream is generated, the system performs integrity checks and device adaptation processing to ensure that the control code stream conforms to the communication protocol and parsing logic required by the injection device. For example, for certain types of programmable syringe systems, control commands must be transmitted in a specific data frame structure, with each frame containing a frame header flag, control type identifier, checksum, and end-of-frame marker. The system automatically adapts the code stream format according to different device types and adds redundant bits and pads the checksum to ensure that no execution abnormalities occur during data transmission due to bit errors. In addition, the system performs a pre-load test run before the injection device is started, that is, inputting the control code stream into the device and performing a simulated run without actually propelling liquid to confirm that the rate and duration of each control period are correctly identified and there are no logical conflicts or delays.

[0148] After all verification and adaptation steps are completed, the injection control commands are officially imported into the program-controlled injector system. Upon receiving the commands, the system will strictly execute the injection operation according to the sequence of injection initiation rate, speed adjustment information, maintenance rate, and termination time set in the control code stream, adjusting the injection rate in real time to ensure that the administration of oxytocin injection throughout the entire treatment cycle highly conforms to the individualized simulation prediction target. Through this method, doctors can execute complex drug administration strategies without manually adjusting the equipment, ensuring both rapid onset of drug concentration in uterine tissue and avoiding excessive drug concentration in the system circulation, achieving the dual goals of improving efficacy and reducing risk.

[0149] In summary, this step not only effectively connects the PBPK simulation analysis results with clinical injection equipment, but also achieves fine-grained control of individualized drug delivery strategies through high-resolution, multi-segment injection behavior modeling and control command encoding.

[0150] Step S106: The injection control command is imported into the syringe system with program control function, and the precise administration of oxytocin injection is performed.

[0151] In step S106, the system imports the simulated, filtered, and formatted injection control instructions into the syringe system with program control capabilities, and executes the actual administration of oxytocin injection according to these instructions. This process marks the formal transition of the invention from the model analysis stage to the physical execution stage. Its core technology lies in ensuring that the control logic is accurately transmitted from the data layer to the hardware layer, and maintaining stability, accuracy, and traceability during injection execution. This step is not only crucial for achieving precise drug administration but also a key step in ensuring the system's clinical usability and the safety and compliance of medical devices.

[0152] Before system deployment, it should be ensured that the syringe device has programmable control functionality. Such devices typically include high-precision micro-infusion pumps, intelligent infusion pumps, or embedded injection modules integrated into anesthesia workstations and obstetric monitoring systems. Their hardware structure should at least include a motor drive module, an injection propulsion mechanism, a control processor, a communication interface, and feedback sensors. The control processor is responsible for parsing injection control commands and scheduling injection actions in real time. The communication interface supports data transmission methods such as USB, RS232, Bluetooth BLE, or CAN bus. Feedback sensors are used to monitor key parameters such as injection pressure, tubing patency, and residual fluid volume in real time.

[0153] In practice, the control command file (usually in JSON, XML, or a specific communication protocol format) generated in the previous step is first uploaded to the syringe system via a computing terminal. The control command file must include the injection initiation rate, the function expression or control curve for the rate variation process, the maintenance rate, the total dosing time, and the emergency interruption threshold. Upon receiving the command, the syringe system must perform integrity verification and parameter validity checks. For example, before receiving the command, the system will automatically determine whether the injection rate exceeds the device's maximum infusion capacity; if the set total dosing time exceeds the maximum support time calculated from the device's cartridge capacity, it will provide a prompt or automatically correct it. Furthermore, the system will perform syntax parsing on key fields such as units (mU / min, μg / h), control node time points, and rate steps to ensure the time-rate mapping of the injection control curve is complete and accurate.

[0154] After the control parameters are loaded, the syringe system enters the pre-operation state. At this time, the system will prompt medical personnel to confirm the operation, including drug loading, tubing venting, interface sealing, and recipient confirmation. The system can also verify that the drug used is oxytocin injection through a drug identification chip or tag identification mechanism to ensure that the instruction matches the type of drug. The device should also automatically detect the motor status, piston position, and remaining dose at the starting point, and will not execute the main injection process until all initial hardware conditions are met.

[0155] Once the drug administration process is officially initiated, the system will dynamically adjust the injection rate with a set time resolution. For example, if the control command is set to linearly increase from 2.0 mU / min to 8.0 mU / min within the first 5 minutes, the system will incrementally increase the rate in each control cycle (e.g., 1 second) to drive the motor to deliver the corresponding dose, while simultaneously calculating the current total injection volume, remaining dose, and rate change trend in real time. If the drug administration parameters are set to a "segmented constant rate + maintenance injection" mode, the system will execute at different fixed rates within each set interval, maintaining a constant output after entering the maintenance phase until the termination time or the uterine contraction response threshold is reached.

[0156] During execution, the syringe system should be linked in real time with external monitoring equipment. Especially during labor induction or labor progression, the system can connect to a uterine contraction pressure monitor or fetal heart rate monitor to receive physiological signals such as the frequency, intensity, and duration of uterine contractions in real time, and dynamically assess the current injection strategy through built-in control logic. For example, if excessively strong or rapid contractions are detected, the system should immediately pause injection, trigger an emergency stop procedure, and send an alarm signal; if the contraction frequency is below the lower limit of the target range for a set time period, the system may consider increasing the maintenance rate or reloading the optimized dosing regimen. This linkage process can achieve closed-loop dynamic control through the feedback closed-loop mechanism preset in this invention, further enhancing the system's intelligence and clinical safety.

[0157] At the end of the entire injection cycle, the syringe system will automatically record all data from the execution process, including the execution command number, injection parameters, timestamp, cumulative dose curve, injection rate curve, user operation record, and any abnormal situations (such as interruption, pause, or restart). This data can be exported as a structured document for use in medical record archiving, quality review, postoperative analysis, or future model retraining. To ensure compliance, all data should have digital signatures and access control.

[0158] In summary, the implementation of step S106 is not simply executing command actions, but a comprehensive control system that spans multiple levels, including hardware initialization, parameter verification, injection control, real-time feedback, safety interruption, and data archiving.

[0159] Furthermore, the step of importing the injection control command into a syringe system with program control function and performing a precise administration of oxytocin injection includes:

[0160] During the syringe system initialization phase, based on the injection start rate, speed adjustment information, maintenance rate and termination time contained in the injection control command, a target injection behavior scheduling sequence with a timestamp index is generated and stored in the real-time task queue of the syringe controller for precise scheduling of the injection rate execution timing and duration of each stage.

[0161] Before drug administration, the closed-loop monitoring module integrated within the syringe system is activated to collect real-time data on the motion status of the propulsion components, motor response delay, and pipeline pressure. Based on the target injection behavior scheduling sequence, feedforward calibration is performed to dynamically fine-tune the injection rate control parameters for each time period, forming a set of actually executable rate control instructions.

[0162] During the drug administration process, the syringe system drives the propulsion component to advance the injection liquid in a precise speed-controlled manner according to the rate control instruction set. At the same time, the integrated pressure sensor and injection flow detection element collect real-time feedback data. When there is a deviation between the feedback data and the set threshold, the execution rate parameters of subsequent instructions in the task queue are automatically updated.

[0163] After drug administration is completed, an injection behavior execution report is constructed based on the complete execution log and feedback data of each stage stored in the syringe system. The execution report records the injection start time, end time, rate change nodes at each stage, number of feedback corrections, and progress stability indicators. The execution report is uploaded to the central management terminal or doctor's terminal for further treatment evaluation and backtracking correction of control model parameters.

[0164] In the implementation of this invention, to accurately import and execute injection control commands generated based on physiological pharmacokinetic simulation results into the syringe system, thereby achieving individualized and precise dosing of oxytocin injection, a complete execution chain including command scheduling, equipment response, closed-loop control, and data backtracking was designed. This process relies on a syringe system with programmable control capabilities, combined with a time indexing mechanism, feedforward adjustment strategy, closed-loop feedback module, and execution log collection mechanism. This ensures that the injection behavior at each stage not only meets the simulated dosing parameter combination requirements but also dynamically responds to equipment response errors and environmental disturbances that may occur during clinical practice, achieving true "on-demand precision control."

[0165] First, upon receiving the target injection control command, the system controller immediately parses the injection initiation rate, speed adjustment information, maintenance rate, and termination time contained in the command. Based on the set sampling period and execution resolution, it constructs an injection behavior scheduling sequence containing timestamps. This scheduling sequence clearly marks the start time, execution duration, and target rate of each stage of the injection behavior, providing a strict time synchronization reference for subsequent system scheduling. Specifically, the system generates a unified time base using an internal high-precision clock, divides each injection stage into several time segments with fixed execution intervals, and associates each segment with corresponding rate target parameters. This scheduling sequence is ultimately loaded into the real-time task queue within the syringe control unit, awaiting execution.

[0166] Once the syringe system is ready—that is, the medication is loaded, the tubing is connected, and air bubbles have been purged—the system enters the pre-administration calibration phase. In this phase, the system first activates its integrated closed-loop monitoring module to collect real-time operational data such as the initial position of the injection propulsion component, servo motor response delay, current consumption, and initial pressure within the infusion tubing. Combining historical device response data with current feedback parameters, the system performs a round of feedforward control parameter fine-tuning. The goal of this operation is to adaptively correct the target rate parameters for each time period in the injection behavior scheduling sequence, enabling it to overcome rate execution deviations caused by mechanical friction, liquid viscosity, and pump head aging during actual propulsion. The fine-tuned rate parameters are encapsulated into a new rate control instruction set. This instruction set strictly retains the temporal structure characteristics of the original instructions, only finely adjusting the rate values ​​to ensure that the system achieves the required time and dosage accuracy during execution.

[0167] Next, the system officially enters the drug delivery phase. In this phase, the syringe controller drives the propulsion assembly to perform the propulsion action according to the rate control instruction set. Specific execution methods can include a servo motor driving a lead screw to advance the syringe piston, a piezoelectric element controlling a micro-pump to propel the drug solution, or a stepper motor controlling a volumetric pump for injection, among others. During propulsion, the system collects feedback data in real time from pressure sensors, position sensors, and flow detection elements. Feedback parameters include, but are not limited to, current propulsion displacement, propulsion rate, liquid flow rate, motor response time, tubing pressure fluctuations, and fluid viscosity parameters. All data is processed and analyzed within the controller using a fast filter and target threshold comparison mechanism to identify whether the system exceeds limits or experiences abnormal fluctuations at any time.

[0168] During feedback monitoring, if the deviation between the current propulsion rate and the target rate exceeds the allowable tolerance range, or if factors that may affect the accuracy of drug infusion, such as sudden pressure changes or abnormal motor load, are detected, the system will immediately trigger a dynamic adjustment mechanism. This mechanism can update the rate parameters of unexecuted stages in the subsequent scheduling sequence in real time according to the type and degree of deviation, correct their rate target values, extend or shorten the execution time of the current stage, and even, in extreme cases, pause the current propulsion process and wait for manual intervention or recalibration. In this way, the system forms a precise injection control closed loop guided by the target rate, based on real-time feedback for correction, and with feedforward fine-tuning and closed-loop updates working in tandem, effectively ensuring the accuracy and physiological effectiveness of the actual drug infusion volume per unit time.

[0169] Once the injection process is complete—that is, the system propulsion components have executed the initial rate segment, speed adjustment segment, and maintenance rate segment according to the injection control commands, and reached the set termination time—the syringe system enters the post-processing stage. In this stage, the system retrieves the complete injection behavior execution log from the storage module and, combined with feedback data collected by the closed-loop control module during execution, constructs a standardized injection behavior execution report. This report details the following key indicators: injection start time, injection termination time, rate change nodes at each stage and their corresponding execution duration, error range between real-time feedback values ​​and the target rate, pressure fluctuations, number of automatic system corrections, task command queue execution integrity status, and propulsion component response accuracy indicators.

[0170] To ensure that the injection execution report can be used by subsequent medical management or clinical evaluation systems, the syringe system generates the report content in a structured data format and uploads it to the central management terminal or physician terminal via wired or wireless means. This execution report is not only used for retrospective analysis of the efficacy of the current treatment process and verification of dosing accuracy, but also for optimizing the parameter settings and weight updates of subsequent PBPK models, thereby forming a complete closed-loop individualized pharmacokinetic-pharmacodynamic feedback control system, improving the accuracy and individual adaptability of the next simulation and control command generation.

[0171] It should be noted that, in order to achieve the command recognition capability, feedback acquisition capability, closed-loop control capability, and data report generation capability required throughout the entire process, the syringe system needs to have the following hardware and software functions: First, an embedded programmable control unit with a task queue scheduling mechanism and a clock synchronization mechanism, supporting the parsing, storage, and execution of command streams; second, a high-resolution position detection and flow detection module, capable of sampling the propulsion status with sub-second time accuracy and feeding back the actual infusion rate; third, an integrated pressure detection channel, used to detect abnormal resistance or infusion blockage risk within the system; fourth, a remote data upload module supporting digital communication interfaces, including but not limited to BLE, Wi-Fi, USB, and other communication methods; and fifth, a software module with local log generation, integrity verification, and system security protection mechanisms to ensure the traceability of the command execution process and the system's anti-interference capability.

[0172] In summary, the present invention provides a control method with complete closed-loop execution, dynamic feedback correction, feedforward fine-tuning and data tracking capabilities in the step of "introducing injection control commands into a syringe system with program control function and performing precise administration of oxytocin injection solution", which is different from the injection schemes in the prior art that are executed by simple constant speed propulsion.

[0173] Furthermore, the precise dosing method for oxytocin injection based on the PBPK model optimization also includes:

[0174] During drug administration, key physiological response indicators of the target receptor are monitored in real time, and the monitoring data are compared with the model prediction results. When the monitoring results deviate from the model prediction by more than the set tolerance range, the model parameters are updated and the drug administration parameters are re-optimized. Based on this, new injection control instructions are generated to adjust the drug administration behavior of the syringe system.

[0175] In a further embodiment of the present invention, a real-time feedback closed-loop control mechanism is introduced during the oxytocin injection administration process optimized based on the PBPK model. The core purpose of this mechanism is to dynamically compare the differences between the key physiological response indicators of the target receptor and the predicted output of the PBPK model through real-time monitoring. This allows for timely adjustment of model parameters and injection behavior in case of significant deviations, ensuring that the entire administration process remains safe, effective, and individually adapted. This step not only constructs an adaptive closed-loop system between model prediction and actual response in terms of technical logic but also enables the entire method to have self-correcting capabilities for individual differences and physiological dynamics.

[0176] In practical implementation, it is first necessary to clarify which physiological response indicators will be selected as key feedback variables. For the application scenario of oxytocin injection, the most critical feedback variables typically include uterine contraction frequency, uterine contraction intensity, contraction duration, fetal heart rate and its variability, and maternal blood pressure. Among these, uterine contraction frequency and intensity can be dynamically measured through external contraction monitoring devices (such as TOCO pressure sensors) or internal pressure catheters. Fetal heart rate signals are collected by fetal electrocardiogram monitoring devices, and maternal vital signs data (such as blood pressure, heart rate, and blood oxygenation) can be obtained in real time through multi-parameter monitors. All these monitoring devices must have open communication interfaces (such as USB, RS232, BLE, or Wi-Fi) and support data exchange with the central control system.

[0177] After receiving the aforementioned physiological monitoring data, the system needs to compare it item by item with the response results predicted by the PBPK model and its associated PD module. To achieve this, the system will pre-output a series of time-stamped predicted response curves during the model simulation phase, including the predicted uterine contraction frequency curve F. pred (t) Predicted fetal heart rate zone HR pred (t) etc. During the operation phase, the system calculates the difference between the actual measured value at each sampling time (e.g., the current uterine contraction frequency is 4 times / 10 minutes) and the predicted value at the corresponding time, and determines whether a correction needs to be triggered based on the set tolerance range. For example, if the PBPK model predicts that there should be a uterine contraction frequency of 5 times / 10 minutes at the 20th minute, but the actual collected value is consistently lower than 3 times / 10 minutes, it indicates that the current drug efficacy is lower than expected. This may be due to factors such as a sluggish uterine response to the drug, accelerated individual metabolism, or an underestimation of the distribution volume. In this case, the model considers a response deviation to have occurred.

[0178] The system judges deviations according to the set tolerance strategy, which can be based on standards such as absolute error (e.g., ±1 time / 10 minutes), relative error (e.g., deviation exceeding ±20% of the predicted value), or statistical confidence interval (e.g., exceeding the 95% confidence interval). If the error is within the tolerance range, the current injection control command remains unchanged; if it exceeds the tolerance range, the model correction process is initiated immediately. Model correction can be performed in several ways. One method is to perform posterior correction of key parameters in the PBPK model using the Bayesian update algorithm, such as updating the Kp value of the uterine compartment, system clearance rate CL, and distribution volume Vd. Another method is to use a parameter re-estimation mechanism, which involves reconstructing the PBPK model using the current monitoring data as new input, running the same dosing parameters with the new model, and analyzing whether its output matches the current physiological state.

[0179] After parameter correction or model replacement, the system re-executes steps S103 and S104 based on the new model state, i.e., re-evaluating whether the current dosing parameters still meet the triple clinical control criteria of uterine drug concentration, systemic blood drug concentration, and uterine contraction prediction. If not, step S105 is executed to generate new injection control instructions, which are then updated to the injection device in real time via the control interface to replace the original injection plan. Upon receiving the new control instructions, the injection device will smoothly transition to the new injection rate and maintenance strategy, avoiding dose abrupt changes or injection interruptions.

[0180] The entire feedback adjustment process must be completed in a closed loop within the chain of data acquisition, model comparison, deviation identification, parameter correction, and control update, and must possess sufficient response speed and computing power. In engineering implementation, it is recommended to deploy a dual-core processing architecture in the control system, with one core responsible for real-time data acquisition and injection execution, and the other dedicated to model simulation and deviation calculation; alternatively, multi-threaded asynchronous computing logic can be used to update the model in the background while maintaining injection continuity. Furthermore, to ensure data security and medical compliance, all monitoring values, comparison results, triggering conditions, adjustment records, and control command versions should be timestamped and archived locally / in the cloud for easy retrospective analysis and risk auditing.

[0181] Through this mechanism, the present invention constructs an intelligent drug delivery closed-loop system with adaptive adjustment capabilities. Its greatest feature is that it no longer relies on doctors to manually adjust the infusion rate according to uterine contraction response, but allows the PBPK model to dynamically adjust itself based on clinical data feedback, thereby being closer to individual reality, responding to physiological changes more quickly, and more effectively ensuring treatment safety and efficacy.

[0182] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

Claims

1. A method for precise administration of oxytocin injection based on PBPK model optimization, characterized in that, include: Collect individual physiological parameter information of the target receptor; Based on the individual physiological parameter information, an individualized parameter dataset is constructed. Based on the individualized parameter dataset, a physiological pharmacokinetic model suitable for the target receptor is constructed, and the physicochemical properties, tissue distribution parameters, metabolism and excretion parameters of the drug related to oxytocin injection are set to simulate the in vivo concentration change process under different administration conditions. Multiple sets of drug administration parameters, including injection initiation rate, maintenance rate and duration of administration, are input into the physiological pharmacokinetic model. Simulation calculations are performed to obtain the corresponding concentration change curves. Based on these curves, the drug concentration level, duration of administration and predicted uterine contraction response of each set of drug administration parameters in uterine tissue and systemic circulation are evaluated. The target dosing parameter combination that simultaneously meets the following clinical control conditions is selected from the simulation results: the drug concentration in the uterine tissue reaches the set effective concentration, the drug concentration in the systemic circulation does not exceed the set safety limit, the dosing duration is within the set range, and the predicted value of uterine contraction response is within the target range. Based on the selected target dosing parameter combination, an injection control command is generated, which includes the injection initiation rate, variable speed adjustment information, maintenance rate and termination time, and is converted into a format that can be recognized by the program-controlled injection device. The injection control command is imported into a syringe system with program control function, and the precise administration of oxytocin injection is performed. Specifically, based on the individualized parameter dataset, a physiological pharmacokinetic model suitable for the target receptor is constructed, and physicochemical properties, tissue distribution parameters, metabolic and excretion parameters related to oxytocin injection are set to simulate the in vivo concentration changes under different administration conditions, including: Based on the individualized parameter dataset, perform individual parameter standardization processing, including unifying units, normalizing reference values, and imputing missing values ​​for weight, height, liver and kidney function indicators, and placental blood flow parameters, to generate a standardized parameter input vector that fits the modeling format. Using the standardized parameter input vector, a structured physiological model of multiple organ compartments is constructed. The model includes at least six functional compartments, including the uterus, liver, kidneys, systemic circulation, placenta and fetus. Each compartment is represented by a system of differential equations to represent the material migration process, and cross-organ connection paths and blood flow rate weights are established. Based on the organ partitioning model, the drug physicochemical property library of oxytocin injection is called to automatically match the drug partition coefficient, tissue affinity coefficient and metabolic enzyme action pathway, and the metabolic clearance rate and renal excretion constant are calculated in combination with individualized liver and kidney function levels to form a parameterized PBPK model for this target receptor. Based on the parameterized PBPK model, a dynamic physiological simulation scenario is constructed. Multiple sets of simulation conditions are set, including different intravenous injection rates, variable speed strategies, and cyclic stress states. Time-step simulation is executed, and drug concentration distribution data in each organ compartment at each time point are output for subsequent concentration-response relationship evaluation and dosing parameter screening.

2. The method for precise administration of oxytocin injection based on PBPK model optimization according to claim 1, characterized in that, Also includes: During drug administration, key physiological response indicators of the target receptor are monitored in real time, and the monitoring data are compared with the model prediction results. When the monitoring results deviate from the model prediction by more than the set tolerance range, the model parameters are updated and the drug administration parameters are re-optimized. Based on this, new injection control instructions are generated to adjust the drug administration behavior of the syringe system.

3. The method for precise administration of oxytocin injection based on PBPK model optimization according to claim 1, characterized in that, The process involves inputting multiple sets of dosing parameters, including injection initiation rate, maintenance rate, and dosing duration, into the physiological pharmacokinetic model, performing simulation calculations to obtain corresponding concentration change curves, and evaluating the drug concentration levels, dosing duration, and predicted uterine contraction responses for each set of dosing parameters in uterine tissue and systemic circulation. An initial set of dosing parameters including injection initiation rate, maintenance rate and dosing duration is constructed, and a candidate set of dosing parameters is obtained through a perturbation screening mechanism based on drug sensitivity weights; The candidate combination of the dosing parameters is input into the physiological pharmacokinetic model, and individualized weighted perturbations of placental blood flow parameters and liver and kidney function indicators are introduced to obtain a set of concentration change curves of uterine tissue and system circulation. Based on the set of concentration change curves, the predicted value of uterine contraction response is calculated by combining the concentration-effect relationship and using a dynamic function with a time delay factor, forming a three-dimensional pharmacodynamic index matrix that includes the peak concentration of uterine tissue, the systemic circulating exposure area, and the response delay time. Based on the three-dimensional pharmacodynamic index matrix, the advantageous solution is extracted according to the principle of maximizing the concentration level in uterine tissue and minimizing the concentration level in the systemic circulating drug, and combined with the drug administration duration constraint, a target drug administration parameter mapping table is formed. The stability of the target dosing parameter combinations in the target dosing parameter mapping table was verified by simulation within the perturbation range of body weight, placental blood flow parameters and liver and kidney function indicators. The combinations that still meet the requirements of uterine tissue drug concentration not lower than the effective concentration, systemic circulating drug concentration not exceeding the safety limit and uterine contraction response prediction value within the target range under the perturbation condition were retained as the recommended dosing parameter results.

4. The method for precise administration of oxytocin injection based on PBPK model optimization according to claim 1, characterized in that, The target dosing parameter combination selected from the simulation results that simultaneously meets the following clinical control conditions: the drug concentration in uterine tissue reaches the set effective concentration, the drug concentration in the systemic circulation does not exceed the set safety limit, the dosing duration is within the set range, and the predicted value of uterine contraction response is within the target range, including: Extract the numerical sequence of drug concentration changes over time in uterine tissue for each combination of injection initiation rate, maintenance rate and administration duration from the simulation results, and identify the time point at which the set effective concentration is first reached, as the reference time point when the combination begins to exert physiological effects in the target receptor. Based on the reference time point, extract the drug concentration change data in the corresponding system cycle, calculate the cumulative concentration exposure value of the combination within a set time window after the reference time point, and compare it with the set safety limit. Exclude all combinations of injection initiation rate, maintenance rate and administration duration whose cumulative concentration exposure value exceeds the set safety limit. For combinations of injection initiation rate, maintenance rate and administration duration that were not excluded, the curves of the corresponding predicted uterine contraction response over time were further analyzed. Multiple evaluation indicators, including response onset time, peak response amplitude, peak duration and response stability, were extracted to quantitatively assess the physiological response performance of all combinations. Based on the results of quantitative assessment, combined with the judgment results of whether the drug concentration in uterine tissue reaches the set effective concentration, whether the drug concentration in the system circulation does not exceed the set safety limit, whether the duration of administration is within the set range, and whether the predicted value of uterine contraction response is within the target range, the combination of injection initiation rate, maintenance rate and duration of administration that meets all clinical control conditions is selected as the target administration parameter combination.

5. The method for precise administration of oxytocin injection based on PBPK model optimization according to claim 1, characterized in that, The step involves generating injection control instructions based on the selected target dosing parameter combinations. These instructions include injection initiation rate, rate adjustment information, maintenance rate, and termination time, and are converted into a format recognizable by the program-controlled injection device. Before generating the injection control command, based on the concentration change curves corresponding to each parameter in the target dosing parameter combination, the time point when the drug concentration in the uterine tissue caused by the injection initiation rate first reaches the set effective concentration is marked on the simulation time axis, and the concentration change rate and the trend of the predicted value of uterine contraction response from this time point to the termination time are recorded. Based on the trend of concentration change rate and uterine contraction response prediction value, the time period when the fluctuation of concentration change rate and uterine contraction response prediction value is greater than a preset threshold is identified, and within the time period, corresponding variable speed adjustment information is generated according to the direction of concentration change, injection rate change history and predicted response risk level. The variable speed adjustment information includes the specified rate increase or decrease and its execution duration. Based on the injection initiation rate, the maintenance rate, the variable speed adjustment information, and the termination time, a complete injection behavior parameter sequence is constructed, and the injection behavior parameter sequence is format-encoded. The injection rate and its change timing and amplitude at each time period are embedded into a unified control command template and converted into a digital control code stream with multi-segment rate adjustment capability. The digital control code stream is subjected to integrity verification and injection device interface adaptation processing to ensure that after the injection control command is transmitted to the syringe system with program control function, the oxytocin injection operation is executed precisely according to the injection start rate, the speed adjustment information, the maintenance rate and the termination time.

6. The method for precise administration of oxytocin injection based on PBPK model optimization according to claim 1, characterized in that, The step of importing the injection control command into a syringe system with program control function and performing precise administration of oxytocin injection includes: During the syringe system initialization phase, based on the injection start rate, speed adjustment information, maintenance rate and termination time contained in the injection control command, a target injection behavior scheduling sequence with a timestamp index is generated and stored in the real-time task queue of the syringe controller for precise scheduling of the injection rate execution timing and duration of each stage. Before drug administration, the closed-loop monitoring module integrated within the syringe system is activated to collect real-time data on the motion status of the propulsion components, motor response delay, and pipeline pressure. Based on the target injection behavior scheduling sequence, feedforward calibration is performed to dynamically fine-tune the injection rate control parameters for each time period, forming a set of actually executable rate control instructions. During the drug administration process, the syringe system drives the propulsion component to advance the injection liquid in a precise speed-controlled manner according to the rate control instruction set. At the same time, the integrated pressure sensor and injection flow detection element collect real-time feedback data. When there is a deviation between the feedback data and the set threshold, the execution rate parameters of subsequent instructions in the task queue are automatically updated. After drug administration is completed, an injection behavior execution report is constructed based on the complete execution log and feedback data of each stage stored in the syringe system. The execution report records the injection start time, end time, rate change nodes at each stage, number of feedback corrections, and progress stability indicators. The execution report is uploaded to the central management terminal or doctor's terminal for further treatment evaluation and backtracking correction of control model parameters.

Images