Infusion pump automatic control method for realizing intelligent infusion

Abstract

The invention relates to the technical field of automatic control of medical instruments, and discloses an infusion pump automatic control method for achieving intelligent infusion. The method comprises the following steps: constructing an infusion time sequence frame according to a multi-medicine infusion scheme, and establishing a dynamic fluid behavior model for describing infusion, retention and flushing processes of medicine liquid in a pipeline by combining pipeline parameters and medicine liquid characteristics; based on the model, a dynamic infusion path from a liquid medicine container to a blood vessel inlet of a patient is generated for each liquid medicine, continuous automatic control instructions including flow velocity change, direction control and valve switching are planned, and finally an infusion pump is driven to execute operation. According to the method, through dynamic accurate modeling and path planning of the liquid medicine in the pipeline, full-process automatic accurate control over multi-medicine sequential infusion is achieved, medicine cross contamination is avoided, and the dosage of flushing liquid and the infusion interval time are reduced.

Description

Technical Field

[0001] This invention relates to the field of automated control technology for medical devices, specifically to an automated control method for an infusion pump that enables intelligent infusion. Background Technology

[0002] Currently, sequential infusion is commonly used when administering multiple drugs intravenously in clinical practice. Conventional techniques often rely on manual replacement of infusion bags by healthcare professionals or the use of multi-channel infusion pumps with programmed control. Essentially, this involves independently starting, stopping, and managing the flow rate of each infusion channel. The infusion pump system merely acts as an actuator, lacking awareness and modeling of the actual physical state of the medication within the tubing. The system cannot know the precise location, mixing interface morphology, or residual amount of different medications within a shared tubing.

[0003] Existing technical solutions have shortcomings. Due to the lack of precise control over the dynamic behavior of fluids within the pipeline, when switching drugs, the removal of residual drugs can only be estimated by injecting a fixed volume of flushing fluid. This can easily lead to insufficient flushing, causing cross-contamination, or excessive flushing, resulting in waste of fluid and time delays. Furthermore, achieving automated continuous infusion of multiple drugs requires complex pipeline design and significant manual intervention. The system cannot dynamically plan drug delivery paths and generate coordinated control commands based on real-time conditions, resulting in low automation and limitations in safety and efficiency. A technical solution is needed that can accurately sense and control the entire drug delivery process within a complex pipeline network, thereby achieving safe, automated, and continuous multi-drug infusion. Summary of the Invention

[0004] The purpose of this invention is to provide an automated control method for an infusion pump that enables intelligent infusion, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an automated control method for an infusion pump for intelligent infusion, the method comprising: Based on a pre-defined multi-drug infusion protocol, an infusion timing framework containing time stamps and event dependencies is constructed. Based on the aforementioned infusion timing framework, and combined with the structural parameters and fluid characteristics of the infusion pipeline, a dynamic fluid behavior model describing the infusion, retention, and flushing process of the drug solution in the pipeline is established. Based on the dynamic fluid behavior model, a complete dynamic infusion path from the drug container to the patient's blood vessel inlet is generated for each type of drug to be infused; For the complete dynamic infusion path, a series of continuous automated control commands including flow rate changes, direction control, and valve switching actions are calculated and planned; The continuous automated control commands are converted into a sequence of control signals that the infusion pump can execute, driving the infusion pump to perform sequential drug infusion and pipeline flushing operations.

[0006] Preferably, the step of constructing an infusion timing framework containing time stamps and event dependencies based on a pre-defined multi-drug infusion scheme includes: Read and parse infusion protocol data containing multiple drug identifiers, target infusion doses for each drug, infusion rates, and infusion order; Based on infusion protocol data, a time-series dependency graph is established with time as the axis and drug infusion events as nodes. The time-series dependency graph clarifies the start time, duration, and flushing interval between each infusion event and the end time of the previous infusion event. Associate each infusion event node in the timing dependency graph with a corresponding infusion tubing physical segment identifier to generate an infusion timing framework that binds time logic, event logic, and physical spatial structure.

[0007] Preferably, the step of establishing a dynamic fluid behavior model describing the infusion, retention, and flushing process of the drug solution in the pipeline, based on the infusion timing framework and combined with the structural parameters and fluid characteristics of the infusion pipeline, includes: Obtain the inner diameter, length, material conformability, and physical properties of the flushing fluid for the infusion tubing; The duration and flushing interval of each infusion event node in the infusion timing framework are used as time boundary conditions; A set of differential equations based on the conservation of mass and momentum is constructed. This set of differential equations is used to simulate the flow velocity distribution, concentration diffusion process, and mixing and displacement behavior between different drug solutions in a specific physical section of the pipeline under given time boundary conditions and pipeline structure parameters, thereby forming a dynamic fluid behavior model.

[0008] Preferably, the step of generating a complete dynamic infusion path from the drug container to the patient's blood vessel inlet for each type of drug solution based on the dynamic fluid behavior model includes: The identifier and physical property parameters of the drug solution to be infused are input into the dynamic fluid behavior model; Run a complete time series simulation in the model from the time point when the infusion of the drug solution begins to the time point when its target dose has completely entered the patient's blood vessels; The records document the trajectory of the position of the infusion solution in each physical segment of the infusion pipeline over time during the simulation, the movement process of the interface between the front and rear ends of the solution, and the interface interaction process between the solution and the previously residual liquid or subsequent flushing liquid in the pipeline. These records constitute the complete dynamic infusion path.

[0009] Preferably, the step of calculating and planning a series of continuous automated control commands, including flow rate changes, direction control, and valve switching actions, for the complete dynamic infusion path, includes: Analyzing the complete dynamic infusion path, the key time points and corresponding positions of the drug solution's advancement, deceleration, pause, and being propelled by the flushing fluid in the pipeline were identified; Based on the characteristics of the infusion pump's drive mechanism and the opening and closing characteristics of the pipeline valves, the required precise instantaneous target flow rate and duration are calculated for each critical time point and location. Based on the precise instantaneous target flow velocity value and the duration, a suitable set of pump head stepper motor control pulse sequences and valve opening / closing state switching instructions are generated. The pump head stepper motor control pulse sequences and valve opening / closing state switching instructions are closely linked in time and together constitute the continuous automated control instructions.

[0010] Preferably, the step of converting the continuous automated control command into a sequence of control signals executable by the infusion pump includes: The control pulse sequence of the pump head stepper motor is compiled into a pulse width modulation signal code that can be directly parsed by the main control chip of the infusion pump; Map the valve opening / closing state switching command to the high / low level change logic and timetable of the infusion pump input / output port; The pulse width modulation signal encoding and the logic and schedule of high and low level changes of input and output ports are time-synchronized to ensure that the motor drive and valve action are precisely coordinated in time to form the final control signal sequence.

[0011] Preferably, the method further includes, after generating the infusion timing framework, performing a logical completeness verification on the infusion timing framework, including: Check for time conflicts in the infusion timing framework, i.e., whether the planned times of different drug infusion events overlap; Verify that the physical segment identifier of the tubing assigned to each drug infusion event matches the actual connected drug container and infusion tubing topology; Confirm that the flushing interval between all infusion events meets the preset minimum flushing time requirement to ensure that the preceding drug solution is fully removed.

[0012] Preferably, the method further includes real-time status feedback and fine-tuning of commands during the sequential infusion of the drug solution and pipeline flushing operations, including: By installing flow sensors and pressure sensors at key locations in the infusion line, the actual flow rate of the drug solution and the pressure data of the line can be collected in real time. The actual flow velocity and pipeline pressure data are compared with the predicted values ​​of the dynamic fluid behavior model at the corresponding time points, and the deviation value is calculated. If the deviation exceeds the allowable threshold, the parameters of the pulse width modulation signal encoding currently being executed will be dynamically adjusted according to the sign and magnitude of the deviation to compensate and correct the infusion rate of the infusion pump in real time.

[0013] Preferably, the method further includes performing infusion process tracing and model parameter self-learning after a complete multi-drug infusion cycle, including: Record the actual execution of all control signal sequences and the actual feedback data from all sensors throughout the entire infusion cycle; The actual execution and feedback data are compared and analyzed globally with the prior predictions of the infusion timing framework and dynamic fluid behavior model. Based on the results of the global comparative analysis, specific parameters in the dynamic fluid behavior model, such as the pipeline flow resistance coefficient or the drug diffusion coefficient, are adaptively calibrated to improve the accuracy of subsequent infusion process simulation and control.

[0014] Preferably, before constructing the infusion timing framework, the method further includes a safety and compatibility pre-check of the input multi-drug infusion protocols, including: Verify the chemical and physical compatibility data of all drugs in the infusion protocol and mark any continuously infused drug pairs that have incompatibilities. Verify that the planned infusion rate for each drug does not exceed its safe infusion rate range; If any incompatibilities or risks of excessive infusion are detected, subsequent steps will be suspended and a warning message will be generated.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By combining the specific structural parameters of the infusion tubing with the fluid characteristics of the medication, a dynamic fluid behavior model describing the infusion, retention, and flushing processes of the medication is established. The system can calculate and visualize the spatial distribution and interface movement of the medication within the tubing in real time. This allows the flushing volume of the tubing dead space to be dynamically adjusted based on the model calculation results, rather than relying on fixed empirical values. This ensures thorough removal of residual medication while minimizing the amount of flushing fluid used and the infusion interval. Simultaneously, this model provides a basis for accurately calculating the actual dosage and timing of the medication entering the patient's body, especially for long tubing or infusions of high-viscosity medications, compensating for dosage errors caused by transmission delays and adsorption within the tubing.

[0016] Based on a dynamic fluid behavior model, a complete dynamic infusion path from source to patient is generated for each medication. A continuous control command sequence, including flow rate, flow direction, and valve coordination, is planned accordingly, enabling automated control of multiple medications in single or multi-branch pipeline systems. This allows different medications to be delivered sequentially along pre-calculated safety channels. The system automatically controls relevant valves to open or close at precise times to guide the medication flow. By planning brief changes in flow rate and direction control, mixed medications retained at pipeline branches or interfaces can be recovered, ensuring the complete dosage of high-value or highly toxic drugs. It also physically eliminates the risk of unintended drug mixing due to valve switching logic errors or improper timing, achieving near-zero cross-contamination continuous infusion operations. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the working principle of the automated control method for intelligent infusion pumps described in this invention. Figure 2 A flowchart for constructing the infusion timing framework; Figure 3 A flowchart for establishing a dynamic fluid behavior model; Figure 4 A timing control diagram for the multiple valve states of an intelligent infusion pump; Figure 5 A comparison chart showing the calibration effect of the width of the mixing zone at the drug-liquid interface in infusion tubing. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This invention provides an automated control method for an infusion pump to achieve intelligent infusion. The method includes: defining time stamps and event dependencies by constructing an infusion timing framework based on a pre-set multi-drug infusion scheme. Using the infusion timing framework as a foundation, and combining the structural parameters and fluid characteristics of the infusion tubing, a dynamic fluid behavior model is established to describe the infusion, retention, and flushing processes of the medication in the tubing. Based on this model, a complete dynamic infusion path from the medication container to the patient's blood vessel inlet is generated for each medication to be infused. For this path, continuous automated control commands, including flow rate changes, direction control, and valve switching actions, are calculated and planned. These commands are converted into a sequence of control signals executable by the infusion pump, driving the pump to perform sequential infusion of the medication and tubing flushing operations.

[0020] Example 1: See Figure 2 Before constructing the infusion sequence framework, a safety and compatibility pre-check is performed on the input multi-drug infusion protocols. The chemical and physical compatibility data of all drugs in the infusion protocol are verified, and consecutively infused drug pairs with incompatibilities are marked. The planned infusion rate for each drug is checked to ensure it does not exceed its safe infusion rate range. If incompatibilities or over-infusion risks are found, subsequent steps are paused and a warning message is generated. Infusion protocol data containing multiple drug identifiers, target infusion doses, infusion rates, and infusion order is read and parsed. Based on the infusion protocol data, a time-dependent graph is constructed with time as the axis and drug infusion events as nodes. The time-dependent graph clearly defines the start time, duration, and flushing interval between each infusion event and the end time of the previous infusion event. Each infusion event node in the time-dependent graph is associated with a corresponding physical segment identifier of the infusion tubing, generating an infusion sequence framework that binds time logic, event logic, and physical spatial structure.

[0021] In practice, the safety and compatibility of the input multi-drug infusion protocols are pre-checked. The chemical and physical compatibility data of all drugs in the infusion protocol are verified, and consecutively infused drug pairs with incompatibilities are marked. The chemical and physical compatibility data comes from a pre-set drug compatibility database, which contains a list of drug incompatibilities. The pre-check process compares the consecutively infused drug pairs in the infusion protocol with the incompatibilities list. If a incompatibility item is matched, an alert is generated. The planned infusion rate of each drug is checked to see if it exceeds its safe infusion rate range. The safe infusion rate range is extracted from the drug safety database. The planned infusion rate is compared with the upper limit of the safe infusion rate range. When the planned infusion rate is greater than the upper limit of the safe infusion rate range, it is determined to be an over-infusion risk. If incompatibilities or over-infusion risks are found, the subsequent steps are suspended and an alert is generated. The alert includes the specific drug identification, risk type, and recommended measures.

[0022] In some embodiments, infusion protocol data containing multiple drug identifiers, target infusion doses, infusion rates, and sequence numbers is read and parsed. The infusion protocol data is input in a structured format. The parsing process extracts the drug identifier, target infusion dose, infusion rate, and sequence number fields. Based on the infusion protocol data, a time-series dependency graph is constructed with time as the axis and drug infusion events as nodes. The time-series dependency graph clearly defines the start time, duration, and flushing interval between each infusion event and the end time of the previous infusion event. It can be understood that the time-series dependency graph is represented by a directed graph structure, where nodes represent infusion events, edges represent time order and dependencies, and each node's attributes include the start time, duration, and flushing interval. Optionally, the time relationships in the time-series dependency graph are expressed by formulas; for the i-th infusion event, its start time... satisfy: in: It is the start time of the (i-1)th infusion event. It is the duration of the (i-1)th infusion event. It is the flushing interval between the (i-1)th infusion event and the ith infusion event. The characters in the formula have the following meanings: This represents the start time of the i-th input event. This indicates the start time of the (i-1)th input event. This represents the duration of the (i-1)th infusion event. This represents the flushing interval between the (i-1)th infusion event and the ith infusion event.

[0023] In specific implementations, each infusion event node in the timing dependency graph is associated with a corresponding physical segment identifier of the infusion tubing, generating an infusion timing framework that binds temporal logic, event logic, and physical spatial structure. The physical segment identifier of the infusion tubing is obtained from the infusion tubing configuration file, which defines the tubing topology and segment identifier mapping. The association process matches nodes with physical segment identifiers based on the location of the drug container and the tubing path required for the drug infusion event. In some embodiments, a safety and compatibility pre-check step is performed before constructing the infusion timing framework. The pre-check result affects the generation of the infusion timing framework; if the pre-check passes, the infusion timing framework continues to be constructed; if the pre-check fails, the process is interrupted. Optionally, the infusion timing framework is stored in the form of spreadsheets or database records, with stored fields including drug identifier, start time, duration, flushing interval, and tubing physical segment identifier.

[0024] Example 2: See Figure 3 The inner diameter, length, material compliance, and physical properties of the flushing fluid in the infusion tubing were obtained. The duration of each infusion event node and the flushing interval in the infusion timeline were used as time boundary conditions. A system of differential equations based on mass and momentum conservation was constructed to simulate the flow velocity distribution, concentration diffusion process, and mixing and displacement behavior between different drug solutions in specific physical segments of the tubing under given time boundary conditions and tubing structural parameters, thus forming a dynamic fluid behavior model. The identifier and physical properties of the drug solution to be infused were input into the dynamic fluid behavior model, and a complete time-series simulation was run from the start of infusion to the point when the target dose was fully absorbed into the patient's blood vessels. The trajectory of the drug solution's position in each physical segment of the infusion tubing over time, the movement of the drug solution's front and rear interfaces, and the interface interaction between the drug solution and previously residual liquid or subsequent flushing fluid in the tubing were recorded, forming a complete dynamic infusion path.

[0025] In practical implementation, the inner diameter, length, material compliance, and physical property parameters of the flushing fluid are obtained. The inner diameter, length, and material compliance parameters of the infusion pipeline are obtained from the hardware configuration data table of the infusion pump system, and the physical property parameters of the flushing fluid, including density and viscosity, are queried from a pre-set flushing fluid property database. The duration of each infusion event node and the flushing interval in the infusion timing framework are used as time boundary conditions, which define the time length constraints of different drug infusion stages and pipeline flushing stages. A system of differential equations based on the conservation of mass and momentum is constructed. This system of differential equations is used to simulate the flow velocity distribution, concentration diffusion process, and mixing and displacement behavior between different drug interfaces in a specific physical section of the pipeline under given time boundary conditions and pipeline structural parameters, thus forming a dynamic fluid behavior model.

[0026] In some embodiments, the system of differential equations includes terms describing the fully developed laminar flow of a fluid in a circular pipe. For a given homogeneous pipe, the relationship between its volumetric flow rate and pressure gradient can be expressed as: in: This indicates the volumetric flow rate through this section of the pipeline. This indicates the inner diameter of this section of the infusion tubing. This indicates the pressure difference acting at both ends of this section of pipeline. Indicates the dynamic viscosity of a fluid. This indicates the length of the pipeline segment. The concentration diffusion process is described using a numerical model based on the convection-diffusion equation, simulating the axial diffusion and mixing behavior of the drug-liquid interface during flow. The numerical model discretizes the pipeline into multiple micro-segments, and applies the convection-diffusion equation within each micro-segment to calculate the spatiotemporal variation of the drug concentration. The convection term characterizes the transport process of the drug with the flow, while the diffusion term characterizes the molecular diffusion effect caused by the concentration gradient. During the simulation, the axial diffusion behavior of the drug-liquid interface is tracked by monitoring the drug concentration distribution within each micro-segment, while the mixing behavior is evaluated by analyzing the concentration superposition of different drugs in the interface region.

[0027] In practice, the identifier and physical property parameters of the drug solution to be infused are input into a dynamic fluid behavior model. These physical property parameters include drug density and viscosity. A complete time-series simulation is run within the dynamic fluid behavior model, starting from the start of drug infusion and ending at the point when the target dose has fully entered the patient's blood vessels. The simulation process progressively solves a system of differential equations according to the sequence and time boundaries set within the infusion timeframe. The simulation records the trajectory of the drug solution's position across each physical segment of the infusion tubing over time, the movement of the drug solution's front and rear interfaces, and the interface interactions between the drug solution and previously residual liquid or subsequent flushing fluid in the tubing, thus forming a complete dynamic infusion path. It can be understood that the trajectory of the drug solution's position across each physical segment of the infusion tubing over time is stored in the form of a time-space data matrix. In some embodiments, the interface position between the drug solution and the flushing fluid is determined by monitoring the concentration distribution of each component in the simulation domain, with the position where the concentration equals 50% defined as the interface center. Optionally, the dynamic fluid behavior model is solved using the finite volume method for spatial discretization and an explicit time-progression method for transient solution. The simulation records the trajectory of the position of the drug solution in each physical segment of the infusion pipeline over time, the movement process of the interface between the drug solution's front and rear ends, and the interface interaction process between the drug solution and previously residual liquid or subsequent flushing fluid in the pipeline. These records together constitute the complete dynamic infusion path for the drug solution.

[0028] Example 3: Analyzing the complete dynamic infusion path, key time points and corresponding positions of drug propulsion, deceleration, pause, and being propelled by flushing fluid within the pipeline are identified. Based on the characteristics of the infusion pump's drive mechanism and the valve switching characteristics, the required precise instantaneous target flow rate and duration are calculated for each key time point and position. Based on the precise instantaneous target flow rate and duration, a suitable set of pump head stepper motor control pulse sequences and valve opening / closing state switching commands are generated. These sequences are tightly synchronized in time, forming continuous automated control commands. The pump head stepper motor control pulse sequence is compiled into a pulse width modulation (PWM) signal encoding that the infusion pump's main control chip can directly parse. The valve opening / closing state switching commands are mapped to a high / low level change logic and timetable for the infusion pump's input / output ports. The PWM signal encoding and the high / low level change logic and timetable for the input / output ports are time-synchronized to ensure precise time coordination between motor drive and valve action, forming the final control signal sequence.

[0029] In practical implementation, the complete dynamic infusion path is analyzed to identify key time points and corresponding positions of the drug solution's advancement, deceleration, pause, and being propelled by flushing fluid within the pipeline. Key time points are obtained by analyzing inflection points or step points on the trajectory curve of the drug solution's position over time, and corresponding positions are determined by the physical segment identifiers of the pipeline recorded in the complete dynamic infusion path. Based on the characteristics of the infusion pump's drive mechanism and the switching characteristics of the pipeline valves, the required precise instantaneous target flow rate and duration are calculated for each key time point and position. The infusion pump's drive mechanism characteristics include the fluid volume corresponding to a single step of the stepper motor, and the pipeline valve switching characteristics include the delay time required for valve state switching. The calculations combine the distance the drug solution needs to move within a specific pipeline segment with the allowable time window.

[0030] In some embodiments, based on the precise instantaneous target flow rate value and its duration, a suitable set of pump head stepper motor control pulse sequences and valve opening / closing state switching instructions are generated. These sequences are closely time-sequential, together constituting continuous automated control instructions. It can be understood that the pump head stepper motor control pulse sequence defines the pulse frequency, pulse quantity, and direction signal. With precise instantaneous target flow velocity value The relationship is determined by the formula: in: This indicates the pulse frequency used to control the stepper motor. This represents the conversion factor determined by the stepper motor step angle and the pump head mechanical structure. This indicates the precise instantaneous target flow rate value. Optionally, valve opening / closing state switching instructions are organized in list form, with each instruction in the list containing the target valve identifier, the target opening / closing state, and the absolute timestamp of execution.

[0031] In specific implementation, the pump head stepper motor control pulse sequence is compiled into a pulse width modulation signal encoding that can be directly parsed by the infusion pump main control chip. The compilation process converts the pulse frequency and direction signals into timer configuration parameters and digital output levels. The valve opening / closing state switching command is mapped to the high / low level change logic and timetable of the infusion pump input / output ports. The mapping relationship is based on a predefined correspondence table between the valve drive circuit and the input / output port address. The pulse width modulation signal encoding and the high / low level change logic and timetable of the input / output ports are time-synchronized to ensure that the motor drive and valve action are precisely coordinated in time, forming the final control signal sequence. It can be understood that time synchronization calibration is achieved through a unified time base, aligning the trigger time of all control events with reference to the same system clock. In some embodiments, the final control signal sequence is loaded into the memory of the infusion pump controller in the form of an executable task list, and executed sequentially by the task scheduler. Optionally, during the generation of continuous automated control commands, resource conflicts or time overlap conflicts between the pump head stepper motor control pulse sequence and the valve opening / closing state switching command are pre-calculated and checked, and adjustments are made.

[0032] See Figure 4 This is a timing control diagram of multiple valves in an intelligent infusion pump. Different colors represent three valves. The horizontal axis represents "time (minutes)," and the vertical axis represents "valve status (open / closed)." A curve in the "open" range indicates that the valve is open, and a curve in the "closed" range indicates that the valve is closed. The three valves open alternately in chronological order without overlap (to avoid conflicts in multiple drug infusions), conforming to the logical requirements of the "multi-drug infusion timing framework." This diagram is used for timing verification of infusion pump valve control, visually demonstrating the time coordination relationship of valve switching, ensuring the independence and safety of different drug infusion paths (avoiding pipeline cross-contamination), and providing a visual basis for the time synchronization calibration of control signal sequences.

[0033] Example 4: After generating the infusion timing framework, its logical completeness is verified. The framework is checked for time conflicts, specifically overlap in the planned times of different drug infusion events. The physical segment identifiers of the tubing assigned to each drug infusion event are verified to match the actual connected drug containers and infusion tubing topology. The flushing intervals between all infusion events are confirmed to meet the preset minimum flushing time requirement, ensuring that preceding drug solutions are adequately removed. Real-time status feedback and fine-tuning of commands are performed during the sequential drug infusion and tubing flushing operations. Flow sensors and pressure sensors located at key positions in the infusion tubing are used to collect real-time data on the actual drug flow rate and tubing pressure. The actual flow rate and pressure data are compared with the predicted values ​​from the dynamic fluid behavior model at the corresponding time points, and the deviation is calculated. If the deviation exceeds the allowable threshold, the parameters of the currently executing pulse width modulation signal encoding are dynamically adjusted based on the sign and magnitude of the deviation to compensate and correct the infusion pump's infusion rate in real time.

[0034] In practical implementation, after generating the infusion timing framework, its logical completeness is verified. This involves checking for time conflicts, specifically whether the planned times of different drug infusion events overlap. The verification process traverses all infusion event nodes in the framework, comparing the start time and duration of each node to determine if any two infusion events have overlapping time intervals. The physical segment identifiers assigned to each drug infusion event are verified to match the actual connected drug containers and infusion tubing topology. This verification is based on a pre-stored infusion tubing connection configuration table, which defines the fixed connection relationships between each drug container and the physical tubing segment. Finally, the flushing intervals between all infusion events are confirmed to meet the preset minimum flushing time requirement to ensure that preceding drug solutions are fully removed. The preset minimum flushing time requirement is calculated based on the tubing volume and flushing flow rate.

[0035] In some embodiments, the results of the logical integrity verification are presented in the form of a report, which lists all the checks and their pass / fail status. See Table 1 for the results of the time conflict check.

[0036] Table 1: Infusion Event Time Conflict Checklist Infusion event identifier Planned start time Planned end time Conflict event identifier Conflict status Event A T1 T2 none No conflict Event B T3 T4 Event C Time overlap Event C T5 T6 Event B Time overlap Understandably, if any verification fails during the check, the process is paused and a prompt is made to correct the infusion sequence framework. During the sequential infusion of the medication and the flushing of the tubing, real-time status feedback and fine-tuning of instructions are performed. Flow sensors and pressure sensors located at key positions in the infusion tubing collect real-time data on the actual flow rate of the medication and the tubing pressure. The flow sensors and pressure sensors operate at a fixed sampling frequency, and the collected data is input to the controller after analog-to-digital conversion.

[0037] In practice, the actual flow velocity and pipeline pressure data are compared with the predicted values ​​of the dynamic fluid behavior model at the corresponding time points, and the deviation value is calculated. The calculation uses the following formula: in: This represents the relative deviation between the actual measured value and the model's predicted value. This represents the actual flow rate or pipeline pressure data collected by the flow sensor or pressure sensor. This represents the flow rate or pressure value predicted by the dynamic fluid behavior model at the corresponding time point. If the deviation exceeds the allowable threshold, the parameters of the currently executing pulse width modulation signal encoding are dynamically adjusted based on the sign and magnitude of the deviation to compensate and correct the infusion rate of the infusion pump in real time. Optionally, the allowable threshold is set according to the safety tolerance of drug infusion, and different allowable thresholds can be set for different drug solutions or different infusion stages. Adjusting the parameters of the pulse width modulation signal encoding is achieved by modifying the reload value of the timer that generates the pulse width modulation signal, thereby changing the pulse frequency. In some embodiments, real-time status feedback and command fine-tuning is a closed-loop process, and multiple acquisition, comparison, and adjustment operations can be performed within the duration of a single infusion event.

[0038] Example 5: After a complete multi-drug infusion cycle, infusion process tracing and model parameter self-learning are performed. The actual execution of all control signal sequences and the actual feedback data from all sensors throughout the entire infusion cycle are recorded. The actual execution and feedback data are then compared globally with the pre-predicted values ​​from the infusion timing framework and the dynamic fluid behavior model. Based on the results of the global comparison analysis, specific parameters in the dynamic fluid behavior model, such as the pipeline flow resistance coefficient or the drug diffusion coefficient, are adaptively calibrated to improve the accuracy of subsequent infusion process simulation and control.

[0039] In practice, after a complete multi-drug infusion cycle, the actual execution status of all control signal sequences and the actual feedback data from all sensors are recorded throughout the infusion cycle. The actual execution status of the control signal sequences includes the actual output timing and parameters of the pulse width modulation signal encoding, and the actual changes in the high and low levels of the input and output ports. The actual feedback data from all sensors includes the time-series data streams collected by the flow sensor and pressure sensor throughout the infusion cycle. The actual execution status and feedback data are then globally compared with the pre-predicted data from the infusion timing framework and dynamic fluid behavior model. This global comparison aligns planned and actual events on the timeline, comparing planned and actual flow rate curves, planned and actual pressure distributions, and planned and calculated drug-liquid interface positions.

[0040] In some embodiments, based on the results of a global comparative analysis, adaptive calibration is performed on specific parameters in the dynamic fluid behavior model, such as the pipeline flow resistance coefficient or the liquid diffusion coefficient. The goal of adaptive calibration is to make the output of the dynamic fluid behavior model more closely resemble actual observation data. It can be understood that the adaptive calibration process employs optimization algorithms, for example, using minimizing the overall deviation between the model's predicted values ​​and the actual sensor feedback data as the objective function to adjust the adjustable parameters in the dynamic fluid behavior model. Optionally, the parameter adjustment amount used for calibration... Calculated by the formula: in: This represents the relative adjustment factor for a specific parameter. This represents the sequence of actual sensor feedback values ​​observed at multiple time points throughout the entire infusion cycle. This represents the sequence of sensor feedback values ​​predicted by the dynamic fluid behavior model at the corresponding time point. The calculated adjustment factor... This will be applied to pre-correct the corresponding parameters in the dynamic fluid behavior model before the next infusion cycle.

[0041] In practice, a global comparative analysis generates a discrepancy report, listing the average and maximum errors between the predicted and actual values ​​of key indicators. The pipeline flow resistance coefficient is calibrated by back-calculation based on actual pressure and flow velocity data, while the drug diffusion coefficient is adjusted based on the difference between the observed width of the drug-liquid interface mixing zone and the model's predicted width. After adaptive calibration, the updated dynamic fluid behavior model parameters are stored for subsequent simulation and control of the infusion process. In some embodiments, the model parameter self-learning process can be set to occur once after several consecutive infusion cycles to accumulate sufficient data for statistically significant calibration.

[0042] See Figure 5This is a comparison chart of the calibration effect of the mixing zone width at the drug-liquid interface in an infusion line. The red curve represents the "mixing zone width before calibration," and the green curve represents the "mixing zone width after calibration." The trend shows that the mixing zone width decreases with increasing infusion cycle number, but the decrease is more significant after calibration. For the same infusion cycle, the mixing zone width after calibration is much smaller than before calibration (e.g., in the 10th cycle, it is 2.90 mm after calibration, compared to 5.90 mm before calibration), indicating that the model parameter self-learning effectively reduced the mixing range at the drug-liquid interface. This chart is used to verify the calibration effect of a dynamic fluid behavior model: a smaller mixing zone width indicates better interface separation of different drugs in the pipeline, reducing the risk of drug incompatibility and improving the safety and accuracy of multi-drug infusions.

[0043] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automated control method for an infusion pump to achieve intelligent infusion, characterized in that, The method includes: Based on a pre-defined multi-drug infusion protocol, an infusion timing framework containing time stamps and event dependencies is constructed. Based on the aforementioned infusion timing framework, and combined with the structural parameters and fluid characteristics of the infusion pipeline, a dynamic fluid behavior model describing the infusion, retention, and flushing process of the drug solution in the pipeline is established. Based on the dynamic fluid behavior model, a complete dynamic infusion path from the drug container to the patient's blood vessel inlet is generated for each type of drug to be infused; For the complete dynamic infusion path, a series of continuous automated control commands including flow rate changes, direction control, and valve switching actions are calculated and planned; The continuous automated control commands are converted into a sequence of control signals that the infusion pump can execute, driving the infusion pump to perform sequential drug infusion and pipeline flushing operations.

2. The automated control method for an infusion pump for intelligent infusion as described in claim 1, characterized in that, The step involves constructing an infusion timing framework that includes time stamps and event dependencies based on a pre-defined multi-drug infusion protocol, including: Read and parse infusion protocol data containing multiple drug identifiers, target infusion doses for each drug, infusion rates, and infusion order; Based on infusion protocol data, a time-series dependency graph is established with time as the axis and drug infusion events as nodes. The time-series dependency graph clarifies the start time, duration, and flushing interval between each infusion event and the end time of the previous infusion event. Associate each infusion event node in the timing dependency graph with a corresponding infusion tubing physical segment identifier to generate an infusion timing framework that binds time logic, event logic, and physical spatial structure.

3. The automated control method for an infusion pump for intelligent infusion as described in claim 2, characterized in that, Based on the aforementioned infusion timing framework, and combined with the structural parameters and fluid characteristics of the infusion pipeline, a dynamic fluid behavior model describing the infusion, retention, and flushing process of the drug solution in the pipeline is established, including: Obtain the inner diameter, length, material conformability, and physical properties of the flushing fluid for the infusion tubing; The duration and flushing interval of each infusion event node in the infusion timing framework are used as time boundary conditions; A set of differential equations based on the conservation of mass and momentum is constructed. This set of differential equations is used to simulate the flow velocity distribution, concentration diffusion process, and mixing and displacement behavior between different drug solutions in a specific physical section of the pipeline under given time boundary conditions and pipeline structure parameters, thereby forming a dynamic fluid behavior model.

4. The automated control method for an infusion pump for intelligent infusion as described in claim 3, characterized in that, Based on the dynamic fluid behavior model, a complete dynamic infusion path is generated for each type of drug solution to be infused, from the drug solution container to the patient's blood vessel inlet, including: The identifier and physical property parameters of the drug solution to be infused are input into the dynamic fluid behavior model; Run a complete time series simulation in the model from the time point when the infusion of the drug solution begins to the time point when its target dose has completely entered the patient's blood vessels; The records document the trajectory of the position of the infusion solution in each physical segment of the infusion pipeline over time during the simulation, the movement process of the interface between the front and rear ends of the solution, and the interface interaction process between the solution and the previously residual liquid or subsequent flushing liquid in the pipeline. These records constitute the complete dynamic infusion path.

5. The automated control method for an infusion pump for intelligent infusion as described in claim 4, characterized in that, For the complete dynamic infusion path, a series of continuous automated control commands are calculated and planned, including flow rate changes, direction control, and valve switching actions, including: Analyzing the complete dynamic infusion path, the key time points and corresponding positions of the drug solution's advancement, deceleration, pause, and being propelled by the flushing fluid in the pipeline were identified; Based on the characteristics of the infusion pump's drive mechanism and the opening and closing characteristics of the pipeline valves, the required precise instantaneous target flow rate and duration are calculated for each critical time point and location. Based on the precise instantaneous target flow velocity value and the duration, a suitable set of pump head stepper motor control pulse sequences and valve opening / closing state switching instructions are generated. The pump head stepper motor control pulse sequences and valve opening / closing state switching instructions are closely linked in time and together constitute the continuous automated control instructions.

6. The automated control method for an infusion pump for intelligent infusion as described in claim 5, characterized in that, The step of converting the continuous automated control commands into a sequence of control signals executable by the infusion pump includes: The control pulse sequence of the pump head stepper motor is compiled into a pulse width modulation signal code that can be directly parsed by the main control chip of the infusion pump; Map the valve opening / closing state switching command to the high / low level change logic and timetable of the infusion pump input / output port; The pulse width modulation signal encoding and the logic and schedule of high and low level changes of input and output ports are time-synchronized to ensure that the motor drive and valve action are precisely coordinated in time to form the final control signal sequence.

7. The automated control method for an infusion pump for intelligent infusion as described in claim 6, characterized in that, The method further includes, after generating the infusion timing framework, performing a logical completeness verification on the infusion timing framework, including: Check for time conflicts in the infusion timing framework, i.e., whether the planned times of different drug infusion events overlap; Verify that the physical segment identifier of the tubing assigned to each drug infusion event matches the actual connected drug container and infusion tubing topology; Confirm that the flushing interval between all infusion events meets the preset minimum flushing time requirement to ensure that the preceding drug solution is fully removed.

8. The automated control method for an infusion pump for realizing intelligent infusion according to claim 7, characterized in that, The method further includes real-time status feedback and fine-tuning of commands during the sequential infusion of the drug solution and pipeline flushing operations, including: By installing flow sensors and pressure sensors at key locations in the infusion line, the actual flow rate of the drug solution and the pressure data of the line can be collected in real time. The actual flow velocity and pipeline pressure data are compared with the predicted values ​​of the dynamic fluid behavior model at the corresponding time points, and the deviation value is calculated. If the deviation exceeds the allowable threshold, the parameters of the pulse width modulation signal encoding currently being executed will be dynamically adjusted according to the sign and magnitude of the deviation to compensate and correct the infusion rate of the infusion pump in real time.

9. The automated control method for an infusion pump for realizing intelligent infusion according to claim 8, characterized in that, The method also includes performing infusion process tracing and model parameter self-learning after a complete multi-drug infusion cycle, including: Record the actual execution of all control signal sequences and the actual feedback data from all sensors throughout the entire infusion cycle; The actual execution and feedback data are compared and analyzed globally with the prior predictions of the infusion timing framework and dynamic fluid behavior model. Based on the results of the global comparative analysis, specific parameters in the dynamic fluid behavior model, such as the pipeline flow resistance coefficient or the drug diffusion coefficient, are adaptively calibrated to improve the accuracy of subsequent infusion process simulation and control.

10. The automated control method for an infusion pump for intelligent infusion according to claim 1, characterized in that, Before constructing the infusion timing framework, the method also includes a safety and compatibility pre-check of the input multi-drug infusion protocols, including: Verify the chemical and physical compatibility data of all drugs in the infusion protocol and mark any continuously infused drug pairs that have incompatibilities. Verify that the planned infusion rate for each drug does not exceed its safe infusion rate range; If any incompatibilities or risks of excessive infusion are detected, subsequent steps will be suspended and a warning message will be generated.

Images