Dynamic management method and system for nuclear medicine split charging activity

Abstract

The invention provides a dynamic management method and system for nuclear medicine split charging activity. The method comprises the following steps: constructing a multi-dimensional safety dynamic graph; performing time sequence aggregation analysis based on the multi-dimensional security dynamic graph, predicting activity change of each subpackage node, and generating a node activity prediction value; according to the node activity predicted value and the multi-dimensional safety dynamic graph, multi-target analysis is carried out, and a scheduling strategy of sub-packaging tasks is generated and comprises a sub-packaging sequence of each task node, a transportation path of each sub-packaging task and a sub-packaging time window; performing simulation verification on the generated scheduling strategy, correcting an abnormal or potential risk, and generating a corrected scheduling strategy; and taking the corrected scheduling strategy as a control signal to execute a subpackage task, collecting activity, transportation path risk and time information in real time to form a feedback data set, and performing intelligent optimization according to the feedback data set.

Description

Technical Field

[0001] The present invention belongs to the field of dynamic activity management, and particularly relates to a method and system for dynamic management of nuclear drug dispensing activity. Background Art

[0002] Nuclear drug dispensing systems are widely used in the production, distribution, and clinical use of radioactive drugs. Their core task is to rationally, efficiently, and safely dispense high-activity nuclear drug doses into individual containers or usage units according to the needs of patients or departments. Nuclear drugs have strong time dependence, and their activity decays rapidly with time. Therefore, dispensing management not only requires precise control of the dispensed dose but also must consider time window management during the dispensing process to maximize activity utilization and reduce resource waste. At the same time, nuclear drug dispensing also needs to strictly control the radiation exposure of staff, follow radiation safety regulations, and ensure the safety of personnel during the dispensing process. Existing nuclear drug dispensing activity management generally relies on manual experience and rule bases for dispensing task scheduling, usually using the method of single-point activity measurement and static formula calculation, that is, through the activity monitoring data of a small number of dispensing nodes, combined with the exponential decay model to estimate the activity changes in each dispensing stage, and then formulate corresponding dispensing strategies.

[0003] This method has many deficiencies.

[0004] Firstly, the complexity of the dispensing process makes it difficult to accurately restore the actual distribution of activity in the dispensing link through simple single-point measurement and static models. Especially in the case of multiple dispensing nodes, multiple operators, dynamic path selection, and uncertain factors (such as sudden operation delays, equipment failures), the activity distribution shows a non-linear and dynamic change trend, which is difficult for existing methods to accurately reflect in real time. Secondly, the dispensing scheduling under the existing method usually relies on manual or preset heuristic rules, lacking a dynamic balance mechanism for multi-objective factors such as activity, time, and personnel exposure, which easily leads to low efficiency of activity dispensing or an increase in safety risks.

[0005] In addition, due to the lack of dynamic global perception and intelligent decision-making capabilities, the current dispensing system is difficult to effectively respond to sudden situations during the dispensing process, such as abnormally rapid decay of activity, conflicts in personnel operation paths, and failures of bottleneck nodes in scheduling. With the continuous expansion of the scale of nuclear drug use in medical institutions, the complexity of the dispensing process is increasing, and the traditional "static activity prediction + manual scheduling" mode is increasingly difficult to meet the requirements of efficient, safe, and dynamic dispensing activity management.

[0006] There is an urgent need for a new dynamic management system that can achieve real-time dynamic perception of the entire process activity and integrate intelligent optimization of multiple factors such as activity utilization efficiency, safety, and dispensing timeliness, so as to improve the overall efficiency and safety level of nuclear drug dispensing. Summary of the Invention

[0007] The object of the present invention is to propose a method and system for dynamic management of the activity of nuclear medicine dispensing, which can effectively solve the prominent defects of the existing methods in aspects such as insufficient dynamic perception of activity, low intelligence of dispensing scheduling, and lack of ability to cope with complex dispensing environments.

[0008] To achieve the above object, in the first aspect of the present invention, a method for dynamic management of the activity of nuclear medicine dispensing is provided. The method includes the following steps:

[0009] Construct a multi-dimensional safety dynamic graph G=(V, E, M, F), where the node V represents a physical or functional area, the edge E represents a physical transfer path, M represents a set of service attributes of the node and the edge, and F is a reserved physical-activity attribute interface.

[0010] Based on the multi-dimensional safety dynamic graph G=(V, E, M, F), perform temporal aggregation analysis to predict the activity changes of each dispensing node and generate node activity prediction values.

[0011] According to the node activity prediction values and the multi-dimensional safety dynamic graph G=(V, E, M, F), perform multi-objective analysis to generate a scheduling strategy for the dispensing task, including the dispensing order of each task node, the transportation path of each dispensing task, and the dispensing time window.

[0012] Perform simulation verification on the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy.

[0013] Execute the dispensing task with the corrected scheduling strategy as the control signal, collect the activity, transportation path risk, and time information in real time to form a feedback data set, and perform intelligent optimization according to the feedback data set.

[0014] Preferably, the multi-dimensional safety dynamic graph G=(V, E, M, F) specifically includes:

[0015] The node attributes include spatial location, controlled area identifier, manual or automated operation area, and node task carrying capacity.

[0016] The edge attributes include the transportation path between two nodes in the dispensing process, including transportation time consumption, path physical distance, transportation mode, whether the path passes through a controlled area, and a safety risk penalty item.

[0017] The physical-activity coupling interface is embedded in the edge attributes and is used to quantify the relationship between activity attenuation and time loss between nodes.

[0018] Preferably, the safety risk penalty item is used to impose avoidance or scheduling penalties on high-radiation risk paths to improve the radiation safety of the dispensing process.

[0019] Preferably, perform temporal aggregation analysis based on the multi-dimensional security dynamic graph G = (V, E, M, F), predict the activity changes of each packaging node, and generate node activity prediction values, specifically:

[0020] Based on the topological relationship of the multi-dimensional security dynamic graph G = (V, E, M, F), use the historical activity status, physical transportation time, and path attenuation factor of adjacent nodes to complete the activity prediction of node v i at the future time t, expressed as:

[0021]

[0022] where is the node activity, represents the historical activity of adjacent node v j at time t - t ji ; t ji is the transportation time of path e ji ; Θ ji is the physical-activity coupling interface, combining physical distance, transportation time, and environmental impact; w ij is the activity propagation weight obtained by graph structure learning, used to model the activity influence intensity from adjacent node v j to node v i ;

[0023] The multi-dimensional security dynamic graph G = (V, E, M, F) also includes an uncertainty regularization mechanism for estimating the activity confidence interval; the uncertainty regularization mechanism is the confidence interval corresponding to the node, obtained by squaring and summing the physical-activity coupling interfaces corresponding to each node in the neighborhood nodes;

[0024] Combine the node activity and the corresponding confidence interval to form the node activity prediction value.

[0025] Preferably, perform multi-objective analysis based on the node activity prediction value and the multi-dimensional security dynamic graph G = (V, E, M, F), and generate a scheduling strategy for the packaging task, specifically including:

[0026] Perform dynamic adjustment analysis of node attributes on the node activity and the corresponding confidence interval to generate a scheduling priority score;

[0027] Collect the cumulative path security risk and path transportation time from the starting point to the target node of the path from the multi-dimensional security dynamic graph G = (V, E, M, F), and obtain the safety-timeliness comprehensive score through multi-factor comprehensive evaluation;

[0028] Execute the scheduling task according to the scheduling priority score and the safety-timeliness comprehensive score.

[0029] Preferably, performing a scheduling task according to the scheduling priority score and the safety-time efficiency comprehensive score specifically includes:

[0030] Generating an execution order {v i1 →v i2 →v in} for the sub-packaging tasks, and preferentially scheduling the nodes with higher scheduling priority score;

[0031] Selecting the optimal transportation path with the best safety-time efficiency comprehensive score for each task node v i ;

[0032] Allocating a dynamic sub-packaging time window based on the node activity for the sub-packaging tasks, and combining the path transportation time and the node activity attenuation characteristics to ensure that the tasks are executed within the effective activity interval.

[0033] Preferably, simulating and verifying the generated scheduling strategy, correcting anomalies or potential risks, and generating a corrected scheduling strategy specifically includes:

[0034] Performing the following verification on each task node v i :

[0035]

[0036] Wherein, is the simulated activity of node v i ; A min is the lower safety limit of the activity; is the actual transportation time consumed in the simulation, is the transportation time set in S; R pi is the risk value of the selected path in S; R max is the maximum allowable path risk threshold set by the sub-packaging system; is the indicator function for verifying whether there are three types of anomalies;

[0037] When Ψ(v i ) > 0, the system determines that there is a simulation anomaly at this node and enters local correction. According to the simulation feedback, the specific scheduling time or path p i of the sub-packaging task v i is corrected in a small range.

[0038] Preferably, the feedback data set includes: the difference between the actual activity and the node activity in the scheduling strategy, and the deviation index between the actual path risk and the risk of the scheduling plan.

[0039] In the second aspect of the present invention, a dynamic management system for the activity of nuclear medicine sub-packaging is provided, and the system includes:

[0040] Data acquisition and modeling module: used to construct a multi-dimensional security dynamic graph G = (V, E, M, F), where node V represents a physical or functional area, edge E represents a physical transfer path, M represents a set of business attributes of nodes and edges, and F is a reserved physical-activity attribute interface.

[0041] Activity prediction module: used to perform time-series aggregation analysis based on the multi-dimensional security dynamic graph G = (V, E, M, F), predict the activity changes of each sub-packaging node, and generate node activity prediction values.

[0042] Scheduling strategy generation module: used to perform multi-objective analysis based on the node activity prediction values and the multi-dimensional security dynamic graph G = (V, E, M, F), and generate a scheduling strategy for the sub-packaging task, including the sub-packaging order of each task node, the transportation path and sub-packaging time window of each sub-packaging task.

[0043] Simulation verification module: used to simulate and verify the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy.

[0044] Sub-packaging execution module: used to execute the sub-packaging task with the corrected scheduling strategy as a control signal, collect activity, transportation path risk and time information in real time, form a feedback data set, and perform intelligent optimization according to the feedback data set.

[0045] The beneficial technical effects of the present invention are at least as follows:

[0046] In view of the above problems, the present invention proposes an intelligent system for dynamic management of the activity of nuclear medicine sub-packaging, which can effectively solve the prominent defects of the existing methods in terms of insufficient dynamic perception of activity, low intelligence of sub-packaging scheduling, and lack of ability to cope with complex sub-packaging environments. Starting from the system level, the present invention constructs a dynamic activity perception mechanism that runs through the entire sub-packaging process, can grasp the activity distribution status of each node in the sub-packaging link in real time, and breaks through the limitations of traditional single-point measurement and static calculation on the modeling ability of dynamic change processes. On this basis, the present invention designs a set of intelligent optimization mechanisms for multi-objective sub-packaging management, which can achieve a dynamic balance among maximizing activity utilization, minimizing personnel radiation exposure, and optimizing sub-packaging scheduling efficiency, and comprehensively improve the intelligent level and safety reliability of the sub-packaging process. At the same time, the present invention constructs a dynamic scheduling system integrating perception and decision-making, and realizes a two-way closed-loop control of activity distribution prediction and sub-packaging decision-making by linking the activity perception module and the scheduling optimization module in real time, can flexibly cope with complex situations such as environmental fluctuations, personnel scheduling changes, and activity anomalies during the sub-packaging process, and has strong system robustness and adaptability. Through the above innovative design, the present invention can significantly improve the accuracy of activity management, the intelligence of sub-packaging scheduling, and the operation safety during the nuclear medicine sub-packaging process, and fully meet the requirements of dynamic management of nuclear medicine activity in complex sub-packaging scenarios. Description of the Drawings

[0047] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the following drawings without creative efforts.

[0048] Figure 1 It is a flowchart of a method for dynamically managing the activity of nuclear drug dispensing according to the present invention. Specific embodiments

[0049] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0050] As Figure 1 shown, a method for dynamically managing the activity of nuclear drug dispensing provided by an embodiment of the present invention includes:

[0051] S1. Construct a multi-dimensional safety dynamic graph G=(V, E, M, F), where the node V represents a physical or functional area, the edge E represents a physical transfer path, M represents a set of service attributes of the node and the edge, and F is a reserved physical-activity attribute interface.

[0052] Specifically, in the task of dynamically managing the activity of nuclear drug dispensing, in order to enable the system to comprehensively perceive elements such as activity transfer, operation risks, and path physical characteristics in the dispensing process, it is first necessary to construct a system topology graph with spatial, physical, safety, and activity coupling characteristics. For this reason, this step designs a multi-dimensional safety dynamic graph (NDSDG, Nuclear Drug Safety Dynamic Graph) as the basic modeling scheme. The system input is the actual layout data of the nuclear drug dispensing center, including the physical coordinates of the dispensing cabinets, buffer areas, transfer channels, functional zoning (controlled area / uncontrolled area), and historical task flow data. By extracting this information, the dispensing center is modeled as a graph structure G=(V, E, M, F), where the node V represents a physical or functional area, the edge E represents a physical transfer path, M represents a set of service attributes of the node and the edge, and F is a reserved physical-activity attribute interface.

[0053] In terms of node modeling, the system takes each physical dispensing unit (such as dispensing cabinet A, buffer area B, transfer channel C) as a node v in the graph i , each v i is attached with attributes p i (spatial location), r i (controlled area identifier), c i (manual or automated operation area) and li (Node task capacity). For example, node v1 of "Sub-packaging Cabinet A" located in the controlled radiation area has r1 = 1 indicating that it is in the controlled area, and c1 = manually marked to allow only manual operation, and l1 = 3 indicating that it supports sub-packaging 3 batches simultaneously.

[0054] In path modeling, the edge e of the graph ij Describes the transportation path between two nodes in the sub-packaging process, including t ij (Transportation time consumption), d ij (Physical distance of the path), b ij (Transportation mode), and q ij (Whether the path passes through the controlled area). For example, the transportation path e from "Buffer B" to "Sub-packaging Cabinet A" 12 , since it passes through the controlled area, assigns q 12 = 1.

[0055] To enhance the adaptability of the model to personnel safety and the physical environment, an innovative safety risk penalty term R is introduced in the attributes of the path ij , which is used as an important basis for path safety discrimination in subsequent activity perception and scheduling optimization:

[0056] R ij = γ·q ij ·r i

[0057] Where γ is the penalty coefficient in the system, q ij indicates whether the path passes through the controlled area (1 means passing through), and r i indicates whether the starting node v i belongs to the controlled area (1 means the controlled area). This design enables the model to dynamically impose significant strategy avoidance or scheduling penalties on high-radiation-risk paths during subsequent activity management, improving the radiation safety of the sub-packaging process.

[0058] In addition, considering the time sensitivity in the nuclear medicine sub-packaging scenario, the model design reserves a physical-activity coupling interface Θ ij , which is directly embedded in the attribute F of the graph edge to quantify the relationship between activity decay and time loss between nodes:

[0059]

[0060] In the formula, τ is the half-life of the processed nuclear medicine (in minutes), t ij is the historical transportation time consumption, d ij is the path length, δ is the environment sensitivity factor, considering the weighted influence of temperature and humidity in the path on activity decay (e.g., δ is larger in a high-temperature environment). Θ ijAs a decay reference term in the subsequent step of dynamic activity prediction, it solves the problem that ordinary sub-packaging diagram models are difficult to reflect the multiple couplings of "space-time-activity" in the transportation path.

[0061] For example, if passing through path e from "buffer B" (v2) 12 to "sub-packaging cabinet A" (v1), path e 12 has a length of 20 meters, takes 5 minutes, the half-life τ of the processed nuclear medicine is 110 minutes, and the current environmental temperature compensation coefficient δ = 0.02, then:

[0062] Safety penalty R 12 = γ × 1 × 1 = γ;

[0063] Activity decay interface

[0064] Through the above method, graph G not only realizes the physical topology modeling of the sub-packaging system, but also pre-integrates the activity propagation and safety constraint interfaces, providing data support for "activity flow modeling between nodes" and "safety-aware scheduling optimization" in the subsequent steps.

[0065] S2. Based on the multi-dimensional security dynamic graph G = (V, E, M, F), perform temporal aggregation analysis to predict the activity changes of each sub-packaging node and generate node activity prediction values.

[0066] Specifically, based on the multi-dimensional security dynamic graph G = (V, E, M, F) established in step 1, this step focuses on completing the dynamic perception and prediction of activities within the nuclear medicine sub-packaging system. The goal is to construct a spatio-temporal distribution prediction model for the activities of each sub-packaging node within the system based on the topological relationship, physical attributes, and activity propagation interface of graph G, providing accurate activity status input for subsequent scheduling strategies.

[0067] The nuclear medicine sub-packaging system has typical characteristics such as "dynamic flow of activity across nodes", "coupled physical path transportation time", and "strong time sensitivity". Therefore, this step proposes a "physically-guided graph activity prediction mechanism (PE-GAPM)", which takes node v i ∈ V as the prediction target and completes the dynamic inference of the system activity based on the activity propagation interface Θ ij in the graph structure and the historical activity of the node.

[0068] Among them, the model first based on the topological relationship of graph G, uses the historical activity status, physical transportation time, and path decay factor of adjacent nodes to complete the activity prediction of node v i at the future time t. The specific definition is as follows:

[0069]

[0070] Among them, Denote the adjacent node v j At time t - t ji The historical activity; t ji For the path e ji The transportation time; Θ ji Is the activity propagation interface defined in step 1, combining physical distance, transportation time, and environmental impact; w ij Is the activity propagation weight obtained from graph structure learning, used to model v j To v i The intensity of activity influence.

[0071] Through the above formula, the model realizes a cross - node activity dynamic aggregation mechanism based on "transportation path - activity physical attenuation - graph structure", which can capture the activity flow relationship of radioactive drugs across time and space in the dispensing system, conforming to the objective law that the activity of radioactive drugs decays with transportation time and physical environment changes.

[0072] Due to the uncertainty of activity in the dispensing system caused by factors such as environmental fluctuations and transportation path diversity, the present invention further designs a physical - environment - aware activity uncertainty regularization mechanism based on graph structure, specifically for estimating the activity confidence interval. Defined as:

[0073]

[0074] Among them, η is the regularization coefficient, controlling the amplitude of uncertainty; Θ ji For the path e ji The activity propagation interface, whose square term enhances the model's perception of activity uncertainty for "high - attenuation, high - time - consuming" paths.

[0075] For example, when transporting radioactive drugs from "buffer B" (v2) to "dispensing cabinet A" (v1), t 12 = 5 minutes, Θ 12 = 0.5, then And The predicted output Is used as the activity state for subsequent scheduling strategy generation modules to call.

[0076] Through the above mechanism, the model realizes the ability of dynamic activity perception for multiple nodes, multiple paths, and multiple time scales in the dispensing system, outputs a complete system activity distribution prediction map, directly interfaces with the subsequent dispensing scheduling optimization module, and serves the "perception - decision - execution - feedback" integrated activity dynamic management system in the patent.

[0077] S3. According to the predicted values of node activities and the multi - dimensional safety dynamic graph G=(V, E, M, F), conduct multi - objective analysis to generate a scheduling strategy for the dispensing task, including the dispensing order of each task node, the transportation path of each dispensing task, and the dispensing time window.

[0078] Specifically, this step uses the node activity prediction output in step 2 and the system graph G=(V, E, M, F) constructed in step 1 as inputs to generate a multi-objective scheduling strategy for the nuclear medicine dispensing task. The dispensing process requires the system to reasonably arrange the dispensing sequence, transportation path, and time plan of each node under the conditions of dynamic activity change, limited transportation path, and controllable safety risk. Therefore, this step proposes an "Activity-Pathway-Safety Model (APSM)" specifically for generating multi-objective scheduling strategies in the nuclear medicine dispensing scenario.

[0079] The input data includes node activity and its confidence interval the safety levels, transportation time consumption, physical topology relationship, etc. of the nodes and paths in graph G. The system takes the target node {v target} of the dispensing task as a constraint and innovatively designs a set of "activity-risk-time" multi-objective scoring mechanisms for scheduling sequence and path decision-making.

[0080] The scheduling priority scoring function is:

[0081]

[0082] where is the current predicted activity of node v i , is the activity uncertainty, and β is the uncertainty sensitivity coefficient. This scoring mechanism ensures that the scheduling strategy preferentially selects target nodes with high activity and strong prediction stability, meeting the requirements of the nuclear medicine dispensing task for activity controllability.

[0083] On this basis, the system designs a "safety-time comprehensive score" function for the transportation path to make decisions on the transportation routes of the dispensing tasks:

[0084]

[0085] where R pi is the cumulative safety risk of path p i from the starting point to the target node v i (derived from R ij in graph G), T pi is the path transportation time, and γ and λ are the weight factors of safety risk and transportation timeliness respectively. This scoring ensures that the scheduling system dynamically balances the radiation safety and transportation time consumption of the path during the dispensing path planning, and automatically selects efficient and low-risk paths under tasks that are sensitive to activity decay and time-constrained.

[0086] Based on the above two-level scoring mechanism, the system completes the following scheduling tasks:

[0087] Execution order of the sub-packaging task Prioritize the scheduling of U(v i ) nodes with high scores;

[0088] For each task node v i Select the transportation path p with the optimal Score(p i ); i ;

[0089] Allocate a dynamic sub-packaging time window based on , combine the transportation time T pi and the node activity decay characteristics to ensure that the task is executed within the effective activity interval.

[0090] For example, if the current activity prediction of node v1 R of path p1 p1 = 0.8, T p1 = 8 minutes, coefficient β = 0.5, γ = 1, λ = 0.05, then:

[0091] Node priority U(v1) = 50 - 0.5×0.5;

[0092] Path score Score(p1) = e -0.8-0.4 = e -1.2 .

[0093] Through the combined two-level scoring, the system preferentially selects sub-packaging tasks with high activity utilization, safe paths and efficient transportation.

[0094] Finally, the scheduling strategy S output by APSM includes:

[0095] The sub-packaging order of each task node;

[0096] The transportation path p of each sub-packaging task i ;

[0097] Sub-packaging time window, the optimal execution time interval based on activity prediction.

[0098] This step focuses on "strategy generation", provides a complete scheduling plan for the subsequent scheduling execution (step 4), and ensures a high degree of consistency between the input (activity status, graph structure) and the output (scheduling plan), strictly conforming to the "perception - decision - execution" chain process.

[0099] S4. Simulate and verify the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy.

[0100] Specifically, this step focuses on systematically simulating and validating based on the scheduling strategy S generated in step 3, and performs limited strategy corrections for the anomalies or potential risks found during the validation process to ensure the physical feasibility, activity effectiveness, and transportation safety of the sub-packaging plan in the nuclear medicine sub-packaging system. The simulation validation inputs include the complete scheduling strategy S generated in step 3 and the predicted node activity values output in step 2 and the confidence interval and relies on the physical constraints and path attributes of the sub-packaging graph G=(V, E, M, F) established in step 1 for support

[0101] During the simulation process, the system reproduces the sub-packaging process task by task and path by path according to the sub-packaging sequence, sub-packaging path, and time arrangement given in S. By simulating the transportation and operation behaviors of each sub-packaging task node v i its simulated activity is calculated and compared with the predicted activity of v i and safety indicators to check for unreasonable situations such as insufficient activity, task timeout, and high-risk paths

[0102] The system defines a "simulation consistency verification index" to perform the following verification on each task node v i as follows:

[0103]

[0104] where is the simulated activity of node v i , dynamically converted based on the transportation time of path p and the decay interface of Θ i ; A ij is the lower limit of activity safety (e.g., 20 mCi), from industry standards; min is the actual transportation time consumed in the simulation, is the transportation time set in S; R is the risk value of the selected path in S; R pi is the maximum allowable path risk threshold set for the sub-packaging system; max is the indicator function to verify the existence of three types of anomalies (insufficient activity, transportation timeout, excessive path risk). When Ψ(v

[0105] i ) > 0, the system determines that there is a simulation anomaly at this node and enters the "local correction mechanism". According to the simulation feedback, the specific scheduling time of the sub-packaging task v i or the path p i is corrected slightly. For example, if in the simulation ​​The system will advance the task execution time based on the original policy or recommend an alternative path.

[0106] During the simulation phase, the system will also output a simulation verification report at the task level, including information such as whether the activity of the task nodes is compliant, whether the path exceeds the risk threshold, and whether the transportation time meets the requirements of the dispensing activity, for use in the subsequent dispensing execution phase.

[0107] For example, if the simulated activity of the dispensing cabinet A(v1) min is lower than A = 20 mCi, and at the same time the simulated transportation time in minutes exceeds the planned

[0108] minutes, then Ψ(v1) = 2 > 0, and the system marks this task as abnormal in the simulation results. * This step finally outputs the scheduling policy S verified by simulation and locally corrected

[0109] as the input for the dispensing execution in step 5, ensuring that the system comprehensively verifies and guarantees the safety, timeliness, and activity management requirements of S in the physical simulation environment.

[0110] Specifically, this step uses the scheduling policy S corrected by simulation output in step 4 * as the input to guide the nuclear medicine dispensing system to execute the dispensing task in the real environment, and during the execution process, it continuously collects the activity, transportation path risk, and key time information in the dispensing operation to form a systematic feedback data set, which supports the entire process of "activity dynamic management and scheduling execution" of this patent in a closed-loop manner.

[0111] The system first converts the dispensing task sequence, path selection, and dispensing time window in S * into specific dispensing control instructions, which cover the start time i of the dispensing at task node v end time actual transportation path p i and transportation duration etc. The automatic control module of the dispensing system drives the dispensing equipment to complete the actual dispensing, transportation, and handover tasks at each node according to S * .

[0112] Among them, during the execution of the dispensing task, the system synchronously collects the actual activity value i of the dispensing node v Path risk exposure data during transportation and the actual completion time of the task node to form a complete execution feedback data stream.

[0113] To provide a standardized execution feedback mechanism for the system, the system defines the "execution - scheduling activity difference index" Γ(v i ), which measures the true activity and the expected activity in the scheduling strategy S * : The difference is as follows:

[0114]

[0115] Among them, is the actual activity of node v collected when the sub - packaging is completed, i and is the predicted value of the activity input by the scheduling strategy. This difference serves as the basic data for the subsequent feedback link of the system, reflecting the degree of coincidence between the activity prediction and the actual situation during the execution process.

[0116] In addition, the system also collects the "deviation index of the actual path risk and the scheduled plan risk" Ω(v i ) to reflect the risk fluctuation during the execution of the transportation path:

[0117]

[0118] Among them, is the real - time monitored risk of the transportation task, and R pi is the path risk value set by the scheduling in S * .

[0119] For example, when the system finishes the sub - packaging execution of the sub - packaging cabinet A(v1) and records the scheduling set value then Γ(v1) = - 1 mCi; if and the planned risk R p1 = 0.8, then Ω(v1) = 0.05, and the system automatically archives it as standard feedback data.

[0120] Through this step, the sub - packaging system has fully realized the physical execution of the S * strategy and high - quality data feedback, completed the technical closed - loop of "perception - decision - verification - execution - feedback" in the patent design, and ensured the activity utilization rate, safety and system stability of the nuclear medicine sub - packaging system in practical applications.

[0121] The embodiment of the present invention also provides a nuclear medicine sub - packaging activity dynamic management system, and the system includes:

[0122] Data acquisition and modeling module: used to construct a multi-dimensional security dynamic graph G = (V, E, M, F), where the node V represents a physical or functional area, the edge E represents a physical transfer path, M represents a set of business attributes of nodes and edges, and F is a reserved physical-activity attribute interface.

[0123] Activity prediction module: used to perform time-series aggregation analysis based on the multi-dimensional security dynamic graph G = (V, E, M, F), predict the activity changes of each sub-packaging node, and generate node activity prediction values.

[0124] Scheduling strategy generation module: used to perform multi-objective analysis based on the node activity prediction values and the multi-dimensional security dynamic graph G = (V, E, M, F), and generate a scheduling strategy for the sub-packaging task, including the sub-packaging order of each task node, the transportation path and sub-packaging time window of each sub-packaging task.

[0125] Simulation verification module: used to simulate and verify the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy.

[0126] Sub-packaging execution module: used to execute the sub-packaging task with the corrected scheduling strategy as a control signal, collect activity, transportation path risk and time information in real time, form a feedback data set, and perform intelligent optimization according to the feedback data set.

[0127] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0128] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect coupling or communication connection of the devices or units can be in an electrical, mechanical, or other form.

[0129] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0130] Although the embodiments of the present invention have been shown and described, those skilled in the art can understand that: various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and purposes of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A dynamic management method for the activity of nuclear medicine dispensing, characterized in that The method includes the following steps: Construct a multi-dimensional security dynamic graph G = (V, E, M, F), where the nodes V represent physical or functional areas, the edges E represent physical transfer paths, M represents the set of service attributes of nodes and edges, and F is a reserved physical-activity attribute interface; Based on the multi-dimensional security dynamic graph G = (V, E, M, F), perform temporal aggregation analysis to predict the activity changes of each sub-packaging node and generate node activity prediction values; According to the node activity prediction values and the multi-dimensional security dynamic graph G = (V, E, M, F), perform multi-objective analysis to generate a scheduling strategy for the sub-packaging task, including the sub-packaging order of each task node, the transportation path and sub-packaging time window of each sub-packaging task; Perform simulation verification on the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy; Use the corrected scheduling strategy as a control signal to execute the sub-packaging task, collect activity, transportation path risk, and time information in real time to form a feedback data set, and perform intelligent optimization according to the feedback data set.

2. The dynamic management method for the dispensing activity of nuclear medicine according to claim 1, characterized in that, The multi-dimensional security dynamic graph G = (V, E, M, F) specifically includes: Node attributes include spatial location, controlled area identifier, manual or automated operation area, and node task carrying capacity; Edge attributes include the transportation path between two nodes in the sub-packaging process, including transportation time, path physical distance, transportation method, whether the path passes through the controlled area, and security risk penalty items; Physical-activity coupling interface, embedded in the edge attributes, used to quantify the relationship between activity attenuation and time loss between nodes.

3. A method for dynamic management of the activity of nuclear drug dispensing according to claim 2, characterized in that, The security risk penalty item is used to impose avoidance or scheduling penalties on high-radiation risk paths to improve the radiation safety of the sub-packaging process.

4. A dynamic management method for the dispensing activity of nuclear medicine according to claim 2, characterized in that The performing temporal aggregation analysis based on the multi-dimensional security dynamic graph G = (V, E, M, F) to predict the activity changes of each sub-packaging node and generate node activity prediction values is specifically as follows: Based on the topological relationship of the multi-dimensional security dynamic graph G = (V, E, M, F), using the historical activity status of adjacent nodes, physical transportation time, and path attenuation factor, the activity prediction of node v i at the future moment t is represented as: Among them, is the node activity, indicating the historical activity of the adjacent node v j at time t - t ji ; t ji is the transportation time consumption of the path e ji ; Θ ji is the physical-activity coupling interface, combining physical distance, transportation time consumption, and environmental impact; w ij is the activity propagation weight obtained by graph structure learning, used to model the activity influence intensity from the adjacent node v j to the node v i . The multi-dimensional security dynamic graph G = (V, E, M, F) also includes an uncertainty regularization mechanism for estimating the activity confidence interval; the uncertainty regularization mechanism is the confidence interval corresponding to the node, which is obtained by squaring and summing the physical-activity coupling interfaces corresponding to each node in the neighborhood nodes; Combine the node activity and the corresponding confidence interval to form the node activity prediction value.

5. A dynamic management method for the dispensing activity of nuclear medicine according to claim 4, characterized in that The performing multi-objective analysis according to the node activity prediction values and the multi-dimensional security dynamic graph G = (V, E, M, F) to generate a scheduling strategy for the sub-packaging task specifically includes: Perform dynamic adjustment analysis of node attributes on the node activity and the corresponding confidence interval to generate a scheduling priority score; Collect the cumulative path security risk and path transportation time from the starting point to the target node of the path from the multi-dimensional security dynamic graph G = (V, E, M, F), and obtain the comprehensive security-time efficiency score through multi-factor comprehensive evaluation; Execute the scheduling task according to the scheduling priority score and the comprehensive security-time efficiency score.

6. The dynamic management method for the dispensing activity of nuclear medicine according to claim 5, characterized in that, The executing the scheduling task according to the scheduling priority score and the comprehensive security-time efficiency score specifically includes: Execution order of the sub-packaging task Prioritize scheduling nodes with high priority scoring scores; For each task node v i Select the transportation route with the optimal comprehensive safety-time efficiency score; Allocate a dynamic sub-packaging time window based on the node activity for the sub-packaging task, and combine the path transportation time and the node activity attenuation characteristics to ensure that the task is executed within the effective activity interval.

7. A method for dynamically managing the activity of nuclear medicine dispensing according to claim 1, characterized in that Performing simulation verification on the generated scheduling strategy, correcting anomalies or potential risks, and generating a corrected scheduling strategy, specifically including: For each task node v i Perform the following verification: Among them, is the simulation activity of node v i ; A min is the lower safety limit of activity; is the actual transportation time consumed in the simulation, is the transportation time set in S; R pi is the risk value of the selected path in S; R max is the maximum allowable path risk threshold set by the sub-packaging system; is an indicator function to verify the existence of three types of anomalies; When Ψ(v i ) > 0, the system determines that there is a simulation anomaly at this node and enters local correction. According to the simulation feedback, the specific scheduling time or path p i of the sub-packaging task v i is corrected within a small range.

8. A method for dynamically managing the activity of nuclear medicine dispensing according to claim 1, characterized in that, The feedback data set includes: the difference between the actual activity and the node activity in the scheduling strategy, and the deviation index between the actual risk of the path and the risk of the scheduling plan.

9. A dynamic management system for the activity of nuclear medicine dispensing, characterized in that, The system includes: Data acquisition and modeling module: used to construct a multi-dimensional security dynamic graph G=(V, E, M, F), where the node V represents a physical or functional area, the edge E represents a physical transfer path, M represents the set of service attributes of the node and the edge, and F is a reserved physical-activity attribute interface. Activity prediction module: used to perform time-series aggregation analysis based on the multi-dimensional security dynamic graph G=(V, E, M, F), predict the activity changes of each sub-packaging node, and generate node activity prediction values. Scheduling strategy generation module: used to perform multi-objective analysis based on the node activity prediction values and the multi-dimensional security dynamic graph G=(V, E, M, F), and generate a scheduling strategy for the sub-packaging task, including the sub-packaging order of each task node, the transportation path and sub-packaging time window of each sub-packaging task. Simulation verification module: used to perform simulation verification on the generated scheduling strategy, correct anomalies or potential risks, and generate a corrected scheduling strategy. Sub-packaging execution module: used to execute the sub-packaging task with the corrected scheduling strategy as the control signal, collect activity, transportation path risk and time information in real time, form a feedback data set, and perform intelligent optimization according to the feedback data set.

Images