Medical service task scheduling method and device, equipment and medium

Abstract

The application discloses a medical service task scheduling method and device, equipment and medium, relates to the technical field of medical information, and comprises the following steps: analyzing medical service center operation demand and medical service object demand through a BERT intention recognition algorithm, and extracting a medical service scheduling core intention; accordingly, a medical service process model is constructed, node flow transfer rules and multi-mode queuing scheduling strategies are configured, and a deducible scheduling simulation model is formed. Historical operation data is input into the model to deduce a scheduling result, if a preset parameter adjustment trigger condition is met, parameters are adaptively adjusted based on the result by using a reinforcement learning algorithm, a new model is generated and re-deduced, until a result meeting a preset verification condition is output, and a target simulation model is determined. Current operation data is input into the target simulation model, and a scheduling scheme is output, so that intelligent scheduling of medical service tasks, process simulation and dynamic parameter optimization are realized, and operation efficiency and resource configuration rationality are improved.

Description

Technical Field

[0001] This invention relates to the field of medical information technology, and in particular to methods, devices, equipment and media for scheduling medical service tasks. Background Technology

[0002] Currently, hospital diagnosis and treatment and health management services have formed a service system with four core parallel scenarios: outpatient, emergency, inpatient, and physical examination. Each scenario undertakes differentiated medical service functions. Outpatient services are responsible for routine visits, follow-up visits, and chronic disease management; emergency services are responsible for emergency treatment of sudden critical illnesses; inpatient services provide full-cycle diagnosis and treatment and rehabilitation management services; and physical examinations are for health screening and preventive healthcare. The process control, resource allocation, and operational efficiency of each scenario directly affect the overall quality of medical services, patient experience, and hospital operational efficiency. At present, the management and scheduling of the four scenarios mostly rely on manual experience, lacking systematic and intelligent unified modeling and extrapolation methods. This makes it difficult to adapt to the core characteristics and dynamic changes of each scenario. Outpatient services cannot cope with traffic fluctuations and multi-stage collaborative scheduling; emergency services struggle to quickly respond to sudden treatments and priority management; inpatient services cannot efficiently achieve bed allocation and multi-departmental collaborative diagnosis and treatment; and physical examinations struggle to complete full-process scheduling and balanced resource allocation. Overall, there is a lack of a unified extrapolation and intelligent scheduling framework covering all scenarios, resulting in common problems such as poor process connection, unreasonable resource allocation, and low service efficiency in each scenario.

[0003] Existing technologies related to medical service scheduling and simulation have significant limitations. Solutions for single scenarios cannot adapt to the unified management needs of multiple parallel scenarios. They generally suffer from insufficient accuracy in demand identification, rigid and singular task scheduling strategies, static and inflexible model parameters, and a lack of multi-module collaboration mechanisms. They cannot accurately capture operational and service needs based on the characteristics of each scenario, nor can they dynamically adjust model parameters according to changes in traffic, scenario switching, and resource status. Furthermore, they have not built a closed-loop simulation and scheduling system covering the entire process of outpatient, emergency, inpatient, and physical examination. This makes it difficult to achieve multi-objective collaborative optimization of process optimization, resource allocation, and order control, and fails to meet the practical needs of refined, intelligent, and integrated management of medical services, thus hindering the further improvement of the overall efficiency and management level of medical services.

[0004] In summary, how to achieve intelligent scheduling, process simulation, and dynamic parameter optimization of medical service tasks, and improve the operational efficiency and resource allocation rationality of medical services, is a problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for scheduling medical service tasks, enabling intelligent scheduling, process simulation, and dynamic parameter optimization of medical service tasks, thereby improving the operational efficiency and resource allocation rationality of medical services. The specific solution is as follows: Firstly, this application discloses a method for scheduling medical service tasks, including: The BERT intent recognition algorithm is used to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent of medical service scheduling. A medical service process model is constructed based on the core intent of the medical service scheduling, and the node flow rules and multi-mode queuing scheduling strategy of the medical service process model are configured to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively. The historical operational data and scenario data of the medical service center are input into the scheduling simulation model, and the simulation is executed to obtain the scheduling results. When the historical operational data meets the preset parameter adjustment triggering conditions, the parameters of the scheduling simulation model are adaptively adjusted based on the scheduling results and through a reinforcement learning algorithm to obtain a new scheduling simulation model. Then, the process jumps back to the step of inputting the historical operational data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so that the obtained scheduling simulation model is determined as the target simulation model. Input the current operational data of the medical service center into the target simulation model to output a scheduling scheme.

[0006] Optionally, the step of parsing the operational needs of the medical service center and the medical service needs of the medical service recipients using the BERT intent recognition algorithm to obtain the core intent for medical service scheduling includes: Collect information on the operational needs of medical service centers and the medical service needs of medical service recipients; The operational requirements and medical service requirements are preprocessed and feature extracted to obtain scheduling intention features; The BERT deep learning model is used to match keywords of medical service scenarios with the scheduling intent features to obtain the core intent of medical service scheduling. The core intent of medical service scheduling includes any one or more of the following: optimization of clinic / bed resources, adjustment of medical service schedule, traffic prediction, queuing time control, priority scheduling of critically ill patients, allocation of medical staff, equipment scheduling, and follow-up / revisit planning.

[0007] Optionally, the step of constructing a medical service process model based on the core intent of the medical service scheduling, and configuring the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model includes: Based on the core intent of the medical service scheduling, each medical service process node is constructed, and the timing and logical relationship of each medical service process node are defined to obtain the medical service process model of each medical service process node. The node jump and multi-path branch rules of the medical service process model are set, and a multi-mode queuing scheduling strategy is configured based on the first-in-first-out scheduling strategy and the priority scheduling strategy to obtain a deducible scheduling simulation model.

[0008] Optionally, the step of adaptively adjusting the parameters of the scheduling simulation model based on the scheduling results and using a reinforcement learning algorithm to obtain a new scheduling simulation model includes: Based on the scheduling results and the current medical service scenario, and through reinforcement learning algorithms, the node service time distribution, task optimization weight, and queuing strategy coefficient in the scheduling simulation model are dynamically optimized to generate updated model parameters that are adapted to the current operating status of the medical service center. The model is reconstructed based on the updated model parameters to obtain a new scheduling simulation model.

[0009] Optionally, the historical operational data includes the actual change in medical service traffic, the actual utilization rate deviation of consultation rooms / beds, the actual queuing time, and the actual medical service scenario. The preset parameter adjustment trigger condition is at least one of the following conditions: The actual change in the medical service flow is greater than the first preset threshold. The actual utilization rate of the consultation room / bed deviates from the second preset threshold. The actual queuing time is greater than the third preset threshold; The actual medical service scenario represents the state switching of the medical service scenario.

[0010] Optionally, the scheduling results include the predicted change in medical service traffic, the predicted utilization rate deviation of clinics / beds, the predicted queuing time, and the predicted medical service scenario; the adaptive adjustment of the parameters of the scheduling simulation model based on the scheduling results and through a reinforcement learning algorithm includes: Based on the target error between the scheduling result and the historical operation data, the parameters of the scheduling simulation model are adaptively adjusted using a reinforcement learning algorithm. The target error includes a first error between the predicted change in medical service traffic and the actual change in medical service traffic, a second error between the predicted utilization rate of consultation rooms / beds and the actual utilization rate of consultation rooms / beds, a third error between the predicted queuing time and the actual queuing time, and a fourth error between the predicted medical service scenario and the actual medical service scenario.

[0011] Optionally, the medical service task scheduling method further includes: The core intent of the medical service scheduling, the model parameters of the scheduling simulation model, the operational data of the medical service center, the scheduling results, and the scheduling scheme are encrypted to obtain encrypted data; wherein, the scheduling scheme includes any one or more of the following: consultation room / bed allocation scheme, patient examination scheduling, medical staff shift scheduling, equipment allocation scheme, triage path planning, follow-up reminder, and follow-up planning. The encrypted data is saved to a distributed database.

[0012] Secondly, this application discloses a medical service task scheduling device, comprising: The intent recognition module is used to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients through the BERT intent recognition algorithm to obtain the core intent of medical service scheduling. The model building module is used to construct a medical service process model based on the core intent of the medical service scheduling, and configure the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively. The model simulation module is used to input the historical operation data and scenario data of the medical service center into the scheduling simulation model and perform simulation to obtain the scheduling results. The model optimization module is used to adaptively adjust the parameters of the scheduling simulation model based on the scheduling result and through a reinforcement learning algorithm when the historical operation data meets the preset parameter adjustment triggering conditions, so as to obtain a new scheduling simulation model. Then, it jumps back to the step of inputting the historical operation data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so as to determine the obtained scheduling simulation model as the target simulation model. The medical service scheduling module is used to input the current operating data of the medical service center into the target simulation model in order to output a scheduling plan.

[0013] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is used to execute the computer program to implement the steps of the aforementioned disclosed medical service task scheduling method.

[0014] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed medical service task scheduling method.

[0015] The beneficial effects of this application are as follows: This application uses the BERT intent recognition algorithm to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent of medical service scheduling; a medical service process model is constructed based on the core intent of medical service scheduling, and the node flow rules and multi-mode queuing scheduling strategy of the medical service process model are configured to obtain a predictable scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks, and inpatient service tasks, respectively; the historical operation data and scenario data of the medical service center are input into the scheduling simulation model, and the simulation is performed to obtain the scheduling result; when the historical operation data meets the preset parameter adjustment trigger condition, the parameters of the scheduling simulation model are adaptively adjusted based on the scheduling result and through the reinforcement learning algorithm to obtain a new scheduling simulation model, and the process jumps back to the step of inputting the historical operation data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset simulation verification condition is obtained, so as to determine the obtained scheduling simulation model as the target simulation model; the current operation data of the medical service center is input into the target simulation model to output the scheduling scheme.Therefore, this application, by employing the BERT intent recognition algorithm to uniformly analyze the operational needs of medical service centers and the medical service needs of medical service recipients, can accurately extract the core intents for medical service scheduling covering the entire scenario of physical examination, emergency, outpatient, and inpatient care. This effectively improves the accuracy and comprehensiveness of demand identification, avoids misjudgment and disconnect from projection, and significantly enhances the adaptability to diverse medical service scenarios. Based on the core intents, a medical service process model is constructed, including corresponding process nodes for physical examination, emergency, outpatient, and inpatient care. Node flow rules and multi-mode queuing scheduling strategies are configured, enabling standardized and high-fidelity modeling of the entire medical service process. This ensures that the process modeling closely matches the actual medical service scenario, improving the applicability and realism of the scheduling simulation model. By inputting historical operational data and scenario data into the simulation model for projection, historical data can be fully utilized to achieve accurate simulation and prediction of the entire medical service process, providing reliable data for scheduling results. The data provides support and basis; when the parameter adjustment triggering conditions are met, the simulation model parameters are adaptively and dynamically adjusted based on the scheduling results using reinforcement learning algorithms. This enables real-time matching and optimization of model parameters with fluctuations in medical service traffic, scene changes, and resource status, effectively improving the model's adaptability, inference accuracy, and scene adaptability, ensuring that the inference results continuously match the actual medical service operation status. By iteratively optimizing until the inference verification conditions are met, the target simulation model is determined, further ensuring the model's stability, reliability, and optimality. Inputting current operational data into the target simulation model and outputting the scheduling scheme enables intelligent, refined, and efficient scheduling and rational allocation of medical service resources, comprehensively improving the efficiency of medical service process operation, resource utilization, and overall service quality, while shortening the waiting time for service recipients, optimizing medical service order and management efficiency, and realizing the integration, intelligence, and efficiency of medical service scheduling across outpatient, emergency, inpatient, and physical examination scenarios. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] Figure 1 This is a flowchart of a medical service task scheduling method disclosed in this application; Figure 2 This is a schematic diagram of medical service task scheduling in a specific outpatient setting disclosed in this application; Figure 3 This application discloses a specific diagram illustrating the scheduling of medical service tasks in an emergency scenario. Figure 4 This is a schematic diagram of medical service task scheduling in a specific inpatient scenario disclosed in this application; Figure 5 This is a schematic diagram of medical service task scheduling in a specific physical examination scenario disclosed in this application; Figure 6 This is a schematic diagram illustrating the implementation of an intelligent agent engine in a specific outpatient scenario as disclosed in this application. Figure 7 This is a schematic diagram illustrating a specific implementation of an intelligent agent engine in an emergency scenario disclosed in this application. Figure 8 This is a schematic diagram illustrating a specific implementation of an intelligent agent engine in a hospitalization scenario disclosed in this application. Figure 9 This is a schematic diagram illustrating the implementation of an intelligent agent engine in a specific physical examination scenario disclosed in this application. Figure 10 This is a schematic diagram of the structure of a medical service task scheduling device disclosed in this application; Figure 11 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Currently, hospital diagnosis and treatment and health management services have formed a service system with four core parallel scenarios: outpatient, emergency, inpatient, and physical examination. Each scenario undertakes differentiated medical service functions. Outpatient services are responsible for routine visits, follow-up visits, and chronic disease management; emergency services are responsible for emergency treatment of sudden critical illnesses; inpatient services provide full-cycle diagnosis and treatment and rehabilitation management services; and physical examinations are for health screening and preventive healthcare. The process control, resource allocation, and operational efficiency of each scenario directly affect the overall quality of medical services, patient experience, and hospital operational efficiency. At present, the management and scheduling of the four scenarios mostly rely on manual experience, lacking systematic and intelligent unified modeling and extrapolation methods. This makes it difficult to adapt to the core characteristics and dynamic changes of each scenario. Outpatient services cannot cope with traffic fluctuations and multi-stage collaborative scheduling; emergency services struggle to quickly respond to sudden treatments and priority management; inpatient services cannot efficiently achieve bed allocation and multi-departmental collaborative diagnosis and treatment; and physical examinations struggle to complete full-process scheduling and balanced resource allocation. Overall, there is a lack of a unified extrapolation and intelligent scheduling framework covering all scenarios, resulting in common problems such as poor process connection, unreasonable resource allocation, and low service efficiency in each scenario.

[0020] Existing technologies related to medical service scheduling and simulation have significant limitations. Solutions for single scenarios cannot adapt to the unified management needs of multiple parallel scenarios. They generally suffer from insufficient accuracy in demand identification, rigid and singular task scheduling strategies, static and inflexible model parameters, and a lack of multi-module collaboration mechanisms. They cannot accurately capture operational and service needs based on the characteristics of each scenario, nor can they dynamically adjust model parameters according to changes in traffic, scenario switching, and resource status. Furthermore, they have not built a closed-loop simulation and scheduling system covering the entire process of outpatient, emergency, inpatient, and physical examination. This makes it difficult to achieve multi-objective collaborative optimization of process optimization, resource allocation, and order control, and fails to meet the practical needs of refined, intelligent, and integrated management of medical services, thus hindering the further improvement of the overall efficiency and management level of medical services.

[0021] Therefore, this application provides a medical service task scheduling scheme to realize intelligent scheduling, process simulation and dynamic parameter optimization of medical service tasks, thereby improving the operational efficiency of medical services and the rationality of resource allocation.

[0022] See Figure 1 As shown in the figure, this application discloses a medical service task scheduling method, including: Step S11: Use the BERT intent recognition algorithm to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent for medical service scheduling.

[0023] The medical service task scheduling system includes a medical service intent accurate identification module, a medical service full-process task scheduling configuration module, a dynamic parameter tuning and optimization module, a medical service scenario simulation and modeling module, a medical service intelligent optimization and execution module, a medical service data archiving module, and a simulation central control module. These modules achieve bidirectional data interaction and collaborative linkage through a high-speed data exchange bus. Specifically, the medical service intent accurate identification module collects and analyzes operational and service recipient needs and extracts the core scheduling intent; the medical service full-process task scheduling configuration module performs time-series planning, path configuration, and orderly scheduling of various medical service process tasks; and the dynamic parameter tuning and optimization module… The medical service scenario simulation modeling module is used to adaptively adjust model parameters based on real-time operational data and scenario changes. It is used to build a high-fidelity process simulation model covering all scenarios of outpatient, emergency and inpatient services and physical examination. The medical service intelligent optimization execution module is used to complete intelligent optimization tasks such as traffic prediction, resource allocation, path planning and queuing optimization based on the simulation model and demand intention. The medical service data archiving module is used to uniformly store the whole process data, model parameters, scheduling results and historical solutions and ensure data security and traceability. The simulation central control module serves as the core unit of the system to coordinate the operation of each module, issue control commands, integrate data information and output the final scheduling results. During the initialization phase, the simulation central control module starts the medical service task scheduling system and controls each module to complete power-on self-checks and data initialization. The medical service intent accurate identification module loads the preset medical service scenario feature library and terminology library and completes the intent recognition model initialization. The medical service full-process task scheduling configuration module loads the process specifications and default scheduling rules for each scenario, including outpatient, emergency, inpatient, and physical examination, and completes the basic configuration of task nodes and priorities. The dynamic parameter tuning and optimization module loads the simulation modeling parameters, task optimization weight parameters, and scenario adaptation parameters based on historical data and completes the initial parameter configuration. The medical service scenario simulation modeling module loads the basic process structure of each scenario and completes the initial modeling of default nodes, flow rules, and queuing scheduling strategies. The medical service intelligent optimization execution module loads the basic optimization algorithm framework and default optimization target weights. The medical service data archiving module completes database connection, encryption strategy activation, and loading and preprocessing of historical operation data and basic resource data. All modules synchronously feed back the initial configuration information to the simulation central control module, which then confirms that each module is in normal condition, the parameters are valid, and the data is ready, completing the initialization process of the entire system and entering the simulation and scheduling state.

[0024] In this embodiment, the step of parsing the operational needs of the medical service center and the medical service needs of the medical service recipients using the BERT intent recognition algorithm to obtain the core intent for medical service scheduling includes: collecting the operational needs of the medical service center and the medical service needs of the medical service recipients; preprocessing and extracting features from the operational needs and the medical service needs to obtain scheduling intent features; and matching medical service scenario keywords with the scheduling intent features using the BERT deep learning model to obtain the core intent for medical service scheduling. The core intent for medical service scheduling includes any one or more of the following: optimization of consultation room / bed resources, adjustment of medical service scheduling, traffic prediction, queuing time control, priority scheduling for critically ill patients, allocation of medical staff, equipment scheduling, and follow-up / re-visit planning.

[0025] For outpatient scenarios, when using the BERT (Bidirectional Encoder Representations from Transformers) intent recognition algorithm for intent parsing, the algorithm first collects the operational needs of the outpatient medical service center, such as departmental load optimization, simplified consultation process, and improved follow-up management, as well as the consultation needs of medical service recipients, such as general consultation, specialist appointment, chronic disease follow-up, and priority consultation. Then, the above operational and consultation needs are processed by text denoising, standardization, and feature extraction to obtain scheduling intent features that match the outpatient scenario. Then, the BERT deep learning model is used to accurately match outpatient scenario keywords, including registration, triage, consultation, examination, follow-up, and chronic disease management, with the scheduling intent features. Finally, the core intents of outpatient medical service scheduling, such as clinic resource optimization, outpatient scheduling adjustment, consultation flow prediction, queuing time control, medical staff allocation, equipment scheduling, and follow-up planning, are obtained.

[0026] For emergency scenarios, when using the BERT intent recognition algorithm for intent parsing, the system first collects the operational needs of the emergency medical service center, such as departmental load balancing, emergency resource allocation, peak traffic management, and treatment process optimization, as well as the emergency treatment needs of medical service recipients, such as critical care, green channel opening, rapid triage, and priority treatment. Then, the system preprocesses these operational and treatment needs, extracts key information, and transforms feature vectors to obtain scheduling intent features that match the characteristics of emergency care. Next, the BERT deep learning model matches emergency scenario keywords, including critical care, rescue, green channel, triage, emergency equipment, and emergency dispatch, with the scheduling intent features. Finally, it obtains the core intents for emergency medical service scheduling, such as priority scheduling for critically ill patients, emergency scheduling adjustment, emergency traffic prediction, waiting time control, emergency deployment of medical personnel, emergency equipment scheduling, and emergency admission route planning.

[0027] For inpatient scenarios, when using the BERT intent recognition algorithm for intent parsing, the system first collects the operational needs of the inpatient medical service center, such as bed allocation, multi-departmental collaborative diagnosis and treatment, inpatient process control, and discharge follow-up optimization, as well as the inpatient diagnosis and treatment needs of medical service recipients, such as admission arrangements, disease diagnosis and treatment, rehabilitation management, discharge planning, and follow-up reminders. Then, the above-mentioned operational and inpatient needs are cleaned, feature-filtered, and structured to obtain scheduling intent features that fit the entire inpatient management cycle. Then, the BERT deep learning model is used to match inpatient scenario keywords, including beds, diagnosis and treatment, rehabilitation, discharge, follow-up, and multi-departmental collaboration, with the scheduling intent features. Finally, the core intents for inpatient medical service scheduling, such as bed resource optimization, inpatient scheduling adjustment, ward traffic prediction, diagnosis and treatment waiting time control, medical staff allocation, equipment scheduling, and discharge follow-up planning, are obtained.

[0028] For physical examination scenarios, when using the BERT intent recognition algorithm for intent parsing, the system first collects the operational needs of the physical examination medical service center, such as optimizing the physical examination schedule, improving the utilization rate of examination rooms, rationally allocating equipment, and simplifying the physical examination process, as well as the physical examination needs of medical service recipients, such as appointment for physical examination, item examination, final examination report, priority physical examination, and other related needs. Then, the above-mentioned operational needs and physical examination needs are preprocessed, feature extracted, and standardized to obtain scheduling intent features adapted to the entire physical examination process. Then, the BERT deep learning model is used to match keywords of the physical examination scenario, including appointment, check-in, examination, final examination, scheduling, examination room, equipment, etc., with the scheduling intent features, and finally obtain the core intents of physical examination medical service scheduling, such as examination room resource optimization, physical examination schedule adjustment, physical examination traffic prediction, queuing time control, medical staff allocation, equipment scheduling, and triage path planning.

[0029] Step S12: Construct a medical service process model based on the core intent of the medical service scheduling, and configure the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively.

[0030] In this embodiment, the step of constructing a medical service process model based on the core intent of medical service scheduling, and configuring the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model includes: constructing each medical service process node based on the core intent of medical service scheduling, defining the timing and configuring the logical relationship of each medical service process node to obtain the medical service process model of each medical service process node; setting the node jump and multi-path branch rules of the medical service process model, and configuring a multi-mode queuing scheduling strategy based on the first-in-first-out scheduling strategy and the priority scheduling strategy to obtain a deducible scheduling simulation model.

[0031] Based on the core intent of medical service scheduling, medical service process nodes corresponding to outpatient, emergency, inpatient, and physical examination are constructed respectively. The timing of each process node is defined and the logical relationship is configured to form a complete medical service process model. Then, flow rules such as node jump and multi-path branch are set, and a multi-mode queuing scheduling mechanism is constructed based on the first-in-first-out strategy and priority strategy. Finally, a scheduling simulation model that can be adapted to the entire medical scenario is formed.

[0032] For outpatient settings, such as Figure 2 As shown, a process model is constructed based on the intentions of process optimization, follow-up visit management, and traffic control, including nodes such as registration, triage, consultation, examination, payment, medication dispensing, and follow-up visits. Rules such as priority triage for follow-up patients and redirection for abnormal examination results are configured. A queuing strategy of first-in-first-out (FIFO) for general patients and priority for patients with chronic diseases and the elderly is adopted. For emergency scenarios, such as... Figure 3 As shown, a process model is constructed based on intentions such as emergency treatment, green channels, and priority for critically ill patients. This model includes nodes such as arrival, triage, rescue, examination, treatment, referral, and discharge. Rules are configured for critically ill patients to directly enter rescue and for sudden illnesses to force a jump to a higher level. A scheduling strategy prioritizing critically ill patients and dedicated to green channels is adopted. For inpatient scenarios, such as... Figure 4 As shown, a process model is constructed based on the intentions of bed allocation, collaborative diagnosis and treatment, and discharge follow-up, including nodes such as admission assessment, treatment implementation, bed management, rehabilitation nursing, discharge planning, and follow-up. Rules are configured for escalation of disease progression and multi-departmental consultation jumps, and a priority strategy of prioritizing severe cases and dynamically adjusting when beds are scarce is adopted. For physical examination scenarios, such as... Figure 5 As shown, a process model is constructed based on scheduling optimization, diversion control, and path planning, including nodes such as appointment, check-in, departmental inspection, final inspection, report collection, and departure. Rules such as project abnormal re-inspection jump and multi-path parallel inspection are configured, and a mixed queuing strategy of conventional first-in-first-out and priority for elderly and group physical examinations is adopted.

[0033] Step S13: Input the historical operation data and scenario data of the medical service center into the scheduling simulation model and perform the simulation to obtain the scheduling result.

[0034] In this embodiment, the historical operational data includes the actual change in medical service traffic, the actual utilization rate deviation of clinics / beds, the actual queuing time, and the actual medical service scenarios.

[0035] In this embodiment, the scheduling results include the predicted change range of medical service traffic, the predicted utilization rate deviation of consultation rooms / beds, the predicted queuing time, and the predicted medical service scenario.

[0036] The system synchronously inputs historical operational data and corresponding scenario data from the medical service center into the scheduling simulation model. Based on the built-in medical service process logic and queuing scheduling strategy, it performs a full-process simulation and outputs scheduling results including the predicted change range of medical service traffic, the deviation of the predicted utilization rate of clinics / beds, the predicted queuing time, and the predicted medical service scenario. The historical operational data covers four key indicators: the actual change range of medical service traffic, the actual utilization rate deviation of clinics / beds, the actual queuing time, and the actual medical service scenario, providing a real data foundation and scenario basis for the simulation process.

[0037] In outpatient settings, historical data such as patient flow fluctuations, clinic load, waiting times, and general / specialist outpatient data are input into the model to extrapolate outpatient flow, clinic utilization, waiting times, and scenario changes. In emergency settings, data such as patient arrival flow, resuscitation room and clinic utilization, waiting times for treatment, and routine / critical care emergency care are input into the model to extrapolate and predict emergency resources and scenarios. In inpatient settings, data such as inpatient flow, bed occupancy, treatment waiting times, and mild / severe / recovery periods are input into the model to predict bed availability, treatment, and inpatient scenarios. In physical examination settings, data such as examination arrival flow, examination clinic and equipment utilization, queuing times, and individual / group physical examinations are input into the model to extrapolate the examination process, resource utilization, and scenario status.

[0038] Step S14: When the historical operation data meets the preset parameter adjustment triggering conditions, based on the scheduling results, the parameters of the scheduling simulation model are adaptively adjusted through a reinforcement learning algorithm to obtain a new scheduling simulation model. Then, the process jumps back to the step of inputting the historical operation data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so as to determine the obtained scheduling simulation model as the target simulation model.

[0039] In this embodiment, the historical operational data includes the actual change range of medical service traffic, the actual utilization rate deviation of clinics / beds, the actual queuing time, and the actual medical service scenario. The preset parameter adjustment trigger condition is at least one of the following conditions: the actual change range of medical service traffic is greater than a first preset threshold; the actual utilization rate deviation of clinics / beds is greater than a second preset threshold; the actual queuing time is greater than a third preset threshold; and the actual medical service scenario indicates that the medical service scenario has a state switch.

[0040] Historical operational data uniformly includes the actual change range of medical service traffic, the actual utilization rate deviation of clinics / beds, the actual queuing / waiting time, and the actual medical service scenario. The corresponding preset parameter adjustment trigger conditions are at least one of the following: the actual change range of medical service traffic exceeds the first preset threshold, the actual utilization rate deviation of clinics or beds exceeds the second preset threshold, the actual queuing time exceeds the third preset threshold, and the actual medical service scenario undergoes a state change. If any condition is met, the model parameters will automatically start adaptive adjustment. Specifically, for outpatient scenarios, historical operational data includes the fluctuation range of outpatient traffic, deviation of consultation room utilization, patient waiting time, and the switching status of scenarios such as general outpatient / specialist outpatient, first-time / follow-up visits, peak / off-peak. Parameter tuning is triggered when the fluctuation of patient traffic exceeds the threshold, the deviation of consultation room load exceeds the standard, the waiting time exceeds the limit, or the outpatient scenario changes. For emergency scenarios, historical operational data includes the fluctuation range of emergency room admission traffic, deviation of emergency room and consultation room utilization, patient waiting time for treatment, and the switching status of scenarios such as routine emergency / emergency care, off-peak / peak, and routine emergency / critical care. Parameter tuning is triggered when the emergency room traffic suddenly increases beyond the threshold, the utilization rate of medical resources exceeds the standard, the waiting time for treatment exceeds the limit, or the emergency scenario changes. For inpatient scenarios, historical operational data includes fluctuations in inpatient traffic, deviations in bed utilization, waiting times for treatment, and statuses such as routine inpatient / emergency admissions, mild / severe / recovery periods, and ward switching. Parameter adjustments are triggered when bed occupancy fluctuations exceed thresholds, bed utilization deviations exceed standards, waiting times for treatment exceed limits, or inpatient scenarios change. For physical examination scenarios, historical operational data includes fluctuations in physical examination arrival traffic, deviations in the utilization of examination rooms and equipment, waiting times for examinees, and statuses such as group / individual physical examinations, peak / off-peak periods, and routine / specialized physical examinations. Parameter adjustments are triggered when physical examination traffic fluctuations exceed thresholds, examination room and equipment utilization deviations exceed standards, waiting times exceed limits, or physical examination scenarios change.

[0041] In a specific embodiment of model parameter adjustment, the scheduling results include the predicted change range of medical service traffic, the predicted utilization rate deviation of clinics / beds, the predicted queuing time, and the predicted medical service scenario. The adaptive adjustment of the parameters of the scheduling simulation model based on the scheduling results and using a reinforcement learning algorithm includes: based on the target error between the scheduling results and the historical operational data, and adaptively adjusting the parameters of the scheduling simulation model using a reinforcement learning algorithm. The target error includes a first error between the predicted change range of medical service traffic and the actual change range of medical service traffic, a second error between the predicted utilization rate deviation of clinics / beds and the actual utilization rate deviation of clinics / beds, a third error between the predicted queuing time and the actual queuing time, and a fourth error between the predicted medical service scenario and the actual medical service scenario.

[0042] For outpatient scenarios, the scheduling results include the predicted change in outpatient traffic, the deviation of the predicted utilization rate of consultation rooms, the predicted waiting time of patients, and the predicted outpatient scenario type. Based on the target error between the scheduling results and historical operational data, the parameters of the scheduling simulation model are adaptively adjusted through reinforcement learning algorithms. The target error includes the first error between the predicted change in outpatient traffic and the actual change, the second error between the deviation of the predicted utilization rate of consultation rooms and the actual utilization rate, the third error between the predicted waiting time and the actual waiting time, and the fourth error between the predicted outpatient scenario and the actual outpatient scenario.

[0043] For emergency scenarios, the scheduling results include the predicted change in emergency patient flow, the deviation between the predicted utilization rates of emergency rooms and clinics, the predicted waiting time for patients, and the predicted emergency scenario type. Based on the target error between the scheduling results and historical operational data, the parameters of the scheduling simulation model are adaptively adjusted through a reinforcement learning algorithm. The target error includes the first error between the predicted change in emergency patient flow and the actual change, the second error between the predicted deviation in the utilization rates of emergency rooms and clinics and the actual utilization rates, the third error between the predicted waiting time for treatment and the actual waiting time for treatment, and the fourth error between the predicted emergency scenario and the actual emergency scenario.

[0044] For inpatient scenarios, the scheduling results include the predicted change in inpatient traffic, the deviation of predicted bed utilization, the predicted patient waiting time, and the predicted inpatient scenario type. Based on the target error between the scheduling results and historical operational data, the parameters of the scheduling simulation model are adaptively adjusted through a reinforcement learning algorithm. The target error includes the first error between the predicted change in inpatient traffic and the actual change, the second error between the deviation of predicted bed utilization and the actual utilization, the third error between the predicted waiting time and the actual waiting time, and the fourth error between the predicted inpatient scenario and the actual inpatient scenario.

[0045] For the physical examination scenario, the scheduling results include the predicted change in the number of examinees arriving at the examination site, the deviation of the predicted utilization rate of examination rooms and equipment, the predicted waiting time for examinees, and the predicted type of physical examination scenario. Based on the target error between the scheduling results and historical operational data, the parameters of the scheduling simulation model are adaptively adjusted through a reinforcement learning algorithm. The target error includes the first error between the predicted change in the number of examinees arriving at the examination site and the actual change, the second error between the predicted deviation of the utilization rate of examination rooms and equipment and the actual utilization rate, the third error between the predicted waiting time and the actual waiting time, and the fourth error between the predicted physical examination scenario and the actual physical examination scenario.

[0046] In another specific embodiment of model parameter adjustment, the step of adaptively adjusting the parameters of the scheduling simulation model based on the scheduling results and through a reinforcement learning algorithm to obtain a new scheduling simulation model includes: based on the scheduling results and the current medical service scenario, dynamically optimizing the node service time distribution, task optimization weights, and queuing strategy coefficients in the scheduling simulation model through a reinforcement learning algorithm to generate updated model parameters that adapt to the current operating status of the medical service center; and reconstructing the model based on the updated model parameters to obtain a new scheduling simulation model.

[0047] Based on the scheduling results and the current medical service scenario, a reinforcement learning algorithm is used to dynamically and adaptively optimize the core parameters of the scheduling simulation model, such as the node service time distribution, task optimization weight, and queuing strategy coefficient. This generates updated model parameters that match the real-time operational status, and the model is reconstructed based on the updated parameters, resulting in a new scheduling simulation model with stronger adaptability and higher inference accuracy.

[0048] For outpatient scenarios, based on scheduling results such as patient flow fluctuations, departmental load, and queuing time, the service time of nodes such as triage, consultation, and examination is optimized, and the optimization weights of consultation efficiency and waiting time are adjusted to dynamically adapt the queuing strategy coefficients for general / specialist outpatient and initial / follow-up consultation scenarios. For emergency scenarios, based on scheduling results such as sudden surges in emergency patient flow, the proportion of critically ill patients, and treatment time, the service time of nodes such as pre-examination triage, rescue, and examination is optimized, the weight of critical care is increased, and the queuing coefficients of green channels and priority are adjusted. For inpatient scenarios, based on scheduling results such as bed occupancy rate, changes in patient condition, and treatment progress, the timing of nodes such as admission assessment, treatment, and discharge is optimized, and the optimization weights of bed utilization rate and rehabilitation effect are adjusted to dynamically adapt the scheduling parameters to the severity of illness and bed shortage status. For physical examination scenarios, based on scheduling results such as physical examination flow, consultation room utilization rate, and examination queuing, the service time of nodes such as check-in, each examination item, and final examination is optimized, and the optimization weights of equipment utilization rate and waiting time are adjusted to adapt the queuing strategy coefficients for off-peak / peak periods.

[0049] Step S15: Input the current operating data of the medical service center into the target simulation model to output a scheduling scheme.

[0050] like Figure 6 As shown, in outpatient settings, the intelligent agent engine captures patient and management needs through intent recognition, constructs a full-process simulation model, and completes task scheduling and intelligent optimization. It dynamically adjusts parameters based on traffic and load changes and outputs solutions such as optimized patient flow, triage queuing, and follow-up appointment reminders. Figure 7 As shown, in emergency scenarios, the intelligent agent engine quickly perceives emergency and operational needs, establishes a simulation model including green channels and priority mechanisms, executes emergency resource scheduling and critical care pathway planning, adjusts parameters in real time based on surges in patient volume and treatment status, and outputs emergency dispatch, medical and equipment allocation, and admission pathway optimization solutions; for example... Figure 8 As shown, in the inpatient scenario, the intelligent agent engine identifies bed availability, treatment, rehabilitation, and follow-up needs, constructs a full-cycle inpatient simulation model, completes bed allocation, treatment scheduling, and follow-up planning, dynamically adjusts parameters based on changes in patient condition and bed availability fluctuations, and outputs bed allocation, medical staff scheduling, multi-departmental collaboration, and discharge follow-up plans; for example... Figure 9 As shown, in the context of physical examinations, the intelligent body engine analyzes the scheduling and resource optimization needs of physical examinations, establishes a full-process simulation model from appointment to departure, implements examination scheduling and path planning, and adaptively adjusts parameters based on the flow of patients and the utilization rate of clinic equipment, outputting physical examination scheduling, triage paths, and equipment and personnel allocation plans.

[0051] This embodiment further includes: encrypting the core intent of the medical service scheduling, the model parameters of the scheduling simulation model, the operational data of the medical service center, the scheduling results, and the scheduling scheme to obtain encrypted data; wherein, the scheduling scheme includes any one or more of the following: clinic / bed allocation scheme, patient examination scheduling, medical staff scheduling, equipment allocation scheme, triage path planning, follow-up reminder, and follow-up planning; and storing the encrypted data in a distributed database. First, the core intent of the medical service scheduling, the model parameters of the scheduling simulation model, the operational data of the medical service center, the scheduling results, and the scheduling scheme are uniformly processed using an encryption algorithm to form secure encrypted data, and all encrypted data is uniformly stored in a distributed database to achieve secure storage, traceable management, and long-term archiving of the entire process data. The scheduling scheme includes any one or more combinations of clinic / bed allocation scheme, patient examination scheduling, medical staff scheduling, equipment allocation scheme, triage path planning, follow-up reminder, and follow-up planning. Specifically, in outpatient settings, encrypted storage includes data such as outpatient visit process parameters, consultation room allocation and waiting schedule, and follow-up visit reminder plans; in emergency settings, encrypted storage includes data such as emergency resource allocation, critical care treatment pathways, and emergency scheduling and equipment dispatch plans; in inpatient settings, encrypted storage includes data such as bed allocation, treatment schedules, medical staff allocation, and discharge follow-up plans; and in physical examination settings, encrypted storage includes data such as physical examination schedules, triage pathways, and consultation room and equipment utilization plans. Data from each scenario is categorized, archived, securely isolated, and quickly queried and accessed through a distributed database.

[0052] The beneficial effects of this application are as follows: This application uses the BERT intent recognition algorithm to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent of medical service scheduling; a medical service process model is constructed based on the core intent of medical service scheduling, and the node flow rules and multi-mode queuing scheduling strategy of the medical service process model are configured to obtain a predictable scheduling simulation model; wherein, the medical service process nodes of the medical service process model are appointment, check-in, examination, final examination, and departure; the historical operational data of the medical service center is input into the scheduling simulation model, and the simulation is performed to obtain the scheduling result; when the historical operational data meets the preset parameter adjustment trigger condition, the parameters of the scheduling simulation model are adaptively adjusted based on the scheduling result and through a reinforcement learning algorithm to obtain a new scheduling simulation model, and the process jumps back to the step of inputting the historical operational data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset simulation verification condition is obtained, so as to determine the obtained scheduling simulation model as the target simulation model; the current operational data of the medical service center is input into the target simulation model to output the scheduling scheme. Therefore, this application, by employing the BERT intent recognition algorithm to accurately analyze the operational needs of medical service centers and the medical service needs of medical service recipients, can quickly and accurately capture the core intent of medical service scheduling, avoid misjudgment of needs, and ensure that subsequent process modeling and scheduling optimization are aligned with actual business scenarios and user needs. Based on the core intent of medical service scheduling, a medical service process model including standard nodes such as appointment, check-in, examination, final examination, and departure is constructed, and node flow rules and multi-mode queuing scheduling strategies are configured. This forms a predictable scheduling simulation model adapted to medical service business, improving the standardization, flexibility, and simulation fit of medical service process modeling. By inputting historical operational data into the scheduling simulation model to execute the simulation and obtain scheduling results, the entire medical service process can be realized based on real historical data. The simulation and quantification of scheduling status provides objective and reliable data support for adjusting model parameters. Based on the scheduling results and combined with reinforcement learning algorithms, the parameters of the scheduling simulation model are adaptively adjusted. The model parameters can be dynamically optimized according to historical operational data and parameter adjustment triggering conditions, continuously improving the accuracy and adaptability of the simulation model, enabling the model to match different operational statuses and scenario changes in medical service centers. By iteratively deducing and verifying to determine the target simulation model, and then inputting the current operational data into the target simulation model to output the scheduling scheme, the scheduling model can be made more accurate and the scheduling scheme can be generated intelligently. This effectively improves the rationality, efficiency, and resource allocation optimization of medical service task scheduling, improves the operational efficiency and service experience of medical service processes, and at the same time increases the utilization rate of medical resources and reduces operational management costs.

[0053] Furthermore, taking the task of providing physical examination services as an example, this application will be explained accordingly.

[0054] This embodiment provides a health management simulation intelligent agent engine device, which specifically includes an intelligent intent recognition module, a management workflow task scheduling node configuration module, an intelligent agent intelligent real-time parameter adjustment module, a simulation modeling module, a task generation optimization module, a data storage module, and a simulation control module. Each module is bidirectionally connected through a bus to collaboratively form a complete health management simulation system, which is adapted to the operational simulation needs of small and medium-sized health management centers.

[0055] The intelligent intent recognition module adopts a BERT-based deep learning model, integrating a demand collection unit and an intent analysis unit. The demand collection unit collects operational demands (such as "optimize the scheduling of medical services from Monday to Friday and increase the utilization rate of clinics to over 85%) and patient medical service appointment demands (such as "prioritize medical services for the elderly and reduce waiting time") through the health management center's management backend. The intent analysis unit preprocesses and extracts features from the collected demand text, accurately identifies the core intent through the trained BERT model, with an accuracy rate of ≥95%, and transmits the recognition results to the inference control module. The original demand data is stored in the data storage module.

[0056] The workflow task scheduling node configuration module integrates the node configuration unit and the task scheduling unit. The node configuration unit receives instructions from the simulation control module and, in conjunction with the process model of the simulation modeling module, configures the flow rules of the entire medical service process nodes (such as jumping to the final examination node after the examination is completed, and jumping to the re-examination node in case of abnormality), and sets the FIFO queuing scheduling strategy (for ordinary patients) and priority strategy (priority for the elderly and pregnant women). The task scheduling unit, according to the process model and queuing strategy, orderly schedules tasks such as medical service appointment, check-in, examination, final examination, and departure, collects data such as clinic utilization rate and patient queuing time in real time, feeds it back to the simulation control module, and synchronously stores it in the data storage module.

[0057] The intelligent agent's real-time parameter tuning module integrates a parameter acquisition unit, a parameter tuning calculation unit, and a parameter verification unit. The parameter acquisition unit collects real-time data on medical service flow (e.g., the number of medical service visits in the morning increases by 20% compared to the preset value), clinic utilization data (e.g., the internal medicine clinic utilization rate is only 70%), and queuing time data (e.g., the ultrasound examination queuing time is 40 minutes). The parameter tuning calculation unit uses a reinforcement learning parameter tuning algorithm, combined with historical parameter adjustment data, to adjust the ultrasound examination node service time distribution parameter in the simulation model to 15-20 minutes, and adjust the equipment utilization optimization weight from 0.3 to 0.4, with a parameter tuning response time ≤100ms. The parameter verification unit transmits the adjusted parameters to the simulation control module for simulation verification, ensuring that the simulation error is ≤5%.

[0058] The simulation modeling module integrates the process modeling unit and the parameter configuration unit. The process modeling unit constructs a logical process model of the entire medical service process, dividing it into 7 core nodes: appointment, check-in, internal medicine examination, surgical examination, ultrasound examination, final examination, and departure. It defines the temporal relationship of each node (e.g., the first examination is arranged within 30 minutes after check-in) and configures the node flow rules. The parameter configuration unit sets the service time distribution of each node (e.g., 10-15 minutes for internal medicine examination and 15-20 minutes for ultrasound examination) to simulate the fluctuation of real medical service efficiency. After the modeling is completed, the model data is fed back to the simulation control module and stored in the data storage module.

[0059] The task generation and optimization module integrates a task generation unit and an optimization analysis unit. Based on the inference model and intent recognition results, the task generation unit generates four optimization tasks: medical service traffic prediction, medical staff scheduling optimization, ultrasound equipment allocation, and intelligent triage path planning. The optimization analysis unit sets optimization objective functions (maximum equipment utilization and minimum patient waiting time) with weights of 0.4 and 0.6 respectively, executes the optimization process, and after optimization, the ultrasound equipment utilization rate increases to 88%, the average patient waiting time is shortened to 18 minutes, and an optimization report is output. Simultaneously, the module retrieves the last three optimization plans for comparison and feeds back to the inference control module. In addition, this module also provides a package sales strategy optimization function, suggesting increased promotion of elderly medical service packages based on medical service traffic prediction results to improve appointment-to-visit conversion rates.

[0060] The data storage module uses a MySQL cluster distributed database to store medical service process model data, medical service traffic data, clinic utilization data, optimization task data, model parameter data, optimization reports, and historical solution data. It supports real-time data reading, writing, backup, and querying, and uses AES encryption algorithm to ensure data security. The data storage capacity can be expanded to 1000GB to meet the long-term operational data storage needs of the health management center.

[0061] The simulation control module receives data and feedback information transmitted from each module, issues control commands such as modeling, scheduling, optimization, and parameter adjustment, coordinates the collaborative work of each module, integrates simulation data (medical service appointment volume, clinic utilization rate, and queuing time) and optimization results, and outputs them to the health management center's management backend for managers to view and make decisions.

[0062] The following is a detailed description of peak medical service scenarios in health management centers, with specific steps as follows: 1.1) Initialization: The simulation control module starts the health management simulation intelligent agent engine, each module completes initialization, and the intelligent agent intelligent real-time parameter tuning module loads the initial model parameters, including node service time distribution (internal medicine 10-15 minutes, ultrasound 15-20 minutes), optimization weights (equipment utilization rate 0.3, waiting time 0.3, personnel idle rate 0.4), queuing scheduling parameters (priority coefficient 1.2), and the initial parameters are stored in the data storage module; 1.2) Data Acquisition: The intelligent intent recognition module collects the operational requirement "to improve the utilization rate of clinics on Monday mornings and shorten patient waiting time". The simulation modeling module collects medical service process data. The workflow task scheduling node configuration module collects data showing that the utilization rate of internal medicine clinics is 70% and the waiting time for ultrasound examinations is 40 minutes. The data storage module collects data showing that the medical service traffic has increased by 20% compared to the preset value. All data is synchronously transmitted to the simulation control module. 1.3) Parameter adjustment trigger judgment: The simulation control module judges that the change in medical service flow is ≥15% and the queuing time is 40 minutes, which exceeds the preset threshold of 30 minutes. If the parameter adjustment trigger condition is met, the parameter adjustment instruction is sent to the intelligent agent intelligent real-time parameter adjustment module. 1.4) Parameter Adjustment: The intelligent agent's real-time parameter adjustment module receives the parameter adjustment instruction, calls the reinforcement learning parameter adjustment algorithm, and combines historical parameter adjustment data to adjust the service time distribution of ultrasound examination nodes to 12-18 minutes, adjust the equipment utilization optimization weight to 0.4, the waiting time weight to 0.4, the personnel idle rate weight to 0.2, and the priority coefficient to 1.3. 1.5) Parameter verification: The intelligent real-time parameter adjustment module of the agent transmits the adjusted parameters to the simulation control module. Based on the adjusted parameters and the simulation model, the simulation control module performs simulation and verification, which shows that the ultrasound examination queue time is shortened to 25 minutes, the utilization rate of the internal medicine clinic is increased to 86%, and the simulation error is 4%≤5%, which meets the preset requirements. 1.6) Parameter update and storage: After verification, the intelligent agent real-time parameter tuning module 3 updates the model parameters, stores them in the data storage module, and synchronously feeds them back to each related module to ensure that each module uses the latest parameters to carry out its work and completes one parameter adjustment. 1.7) Loop execution: If the medical service simulation is not completed on Monday morning, return to step 2 to continue collecting data. Subsequently, it is observed that the medical service traffic tends to be stable and the parameter adjustment conditions are not triggered until the medical service simulation and optimization tasks are completed in the morning, and the parameter adjustment loop ends.

[0063] The following is a simulation and optimization of providing a week's worth of physical examination services to a health management center. The specific steps are as follows: 2.1) Engine Startup: The simulation control module starts the health management simulation intelligent engine, each module completes initialization, loads preset parameters and configuration information, and the data storage module loads the health management center's physical examination operation data for the past month, as well as the basic information of 5 clinics and 3 ultrasound devices; 2.2) Intent Recognition: The intelligent intent recognition module 1 collects the operational needs of the health management center, namely, "Optimize this week's physical examination schedule, improve the utilization rate of clinics and equipment, shorten the average patient waiting time to within 20 minutes, and improve the appointment conversion rate." Through the BERT intent recognition algorithm, it accurately captures the core intent and transmits it to the inference control module. 2.3) Modeling and Deduction: The modeling and deduction control module sends modeling instructions to the modeling and deduction modeling module. The modeling and deduction modeling module constructs a logical flow model of the entire physical examination process, dividing it into 7 core nodes: appointment, check-in, internal medicine examination, surgical examination, ultrasound examination, final examination, and departure. It defines the temporal relationship of each node, configures node flow rules (such as jumping to the re-examination node if the ultrasound examination is abnormal), sets the FIFO queuing scheduling strategy (for ordinary patients) and priority strategy (priority for the elderly and pregnant women), sets the service time distribution of each node, simulates the fluctuation of real physical examination efficiency, and feeds back the model data to the modeling and deduction control module 7 after the modeling is completed. 2.4) Task Scheduling: Based on the intent recognition results and the inference model, the simulation control module issues scheduling instructions to the management workflow task scheduling node configuration module. This module schedules the entire physical examination process tasks in an orderly manner, collects data such as the utilization rate of each clinic, patient queuing time, and number of physical examination appointments in real time, and feeds them back to the simulation control module. 2.5) Task Generation and Optimization: The simulation control module sends optimization instructions to the task generation and optimization module 5, which generates 6 optimization tasks: prediction of this week's physical examination traffic, optimization of medical staff scheduling, allocation of ultrasound equipment, intelligent triage path planning, dynamic adjustment of resource allocation, and optimization of sales strategy for physical examination packages; sets the optimization objective function and weights (equipment utilization rate 0.3, waiting time 0.3, appointment conversion rate 0.2, staff idle rate 0.2), executes the optimization process, and the optimization results are: increase the number of ultrasound devices from 3 to 4 (during peak hours), adjust the medical staff scheduling (add 2 nurses on Mondays and Wednesdays), plan intelligent triage paths (prioritize internal medicine and ultrasound examinations for the elderly), and suggest increasing the promotion of physical examination packages for middle-aged and young people; outputs an optimization report, retrieves the last two optimization plans for comparison, and feeds back to the simulation control module 7; 2.6) Dynamic parameter tuning: The intelligent real-time parameter tuning module 3 of the intelligent agent dynamically adjusts the model parameters according to the parameter adjustment method in Example 2 based on real-time data such as the physical examination traffic (18% higher than the preset value) and the clinic utilization rate (75% on average) of the first two days of this week and the feedback on the optimization effect, so as to ensure the accuracy of the inference data. 2.7) Data output of simulation: The simulation control module integrates simulation data (daily physical examination appointments this week, utilization rate of each clinic, average patient waiting time) and optimization results, outputs optimization reports and historical plan comparison results, and feeds them back to the health management center management backend; 2.8) Resource optimization implementation: Based on the simulation data and optimization plan, the health management center manager will adjust the physical examination schedule for this week, add one ultrasound device and two nurses during peak hours, and implement intelligent triage path and package sales strategy; 2.9) End of simulation: The simulation and optimization tasks for this week's physical examination operation have been completed. The simulation control module shuts down the engine, saves all relevant data, and the simulation process is over. According to statistics, the average utilization rate of the clinic increased to 88% this week, the average waiting time for patients was shortened to 17 minutes, the appointment conversion rate increased by 10%, and the overall operational efficiency was significantly improved.

[0064] The following is a detailed description of the outpatient service scenario, with the specific steps as follows: 1.1) Initialization Configuration: The simulation central control module starts the outpatient simulation intelligent engine, each module completes initialization self-check, and the outpatient dynamic parameter tuning and optimization module loads the preset initial parameters. The initial parameters are preset based on the outpatient routine operation scenario, diagnosis and treatment guidelines and historical medical data, and mainly cover three types of core parameters: outpatient simulation modeling parameters (medical process sequence, medical priority coefficient, scenario simulation parameters, etc.), task optimization weight parameters (medical efficiency weight, patient waiting time weight, department load balancing weight, etc.), and parameter tuning adaptation parameters (traffic fluctuation adaptation coefficient, scenario difference adaptation parameters, etc.). All initial parameters are synchronously archived to the outpatient data archiving module to ensure that the parameters are traceable and reviewable.

[0065] 1.2) Multi-source data acquisition: The outpatient intent accurate recognition module collects outpatient management needs and visitation scenario information; the outpatient scenario simulation modeling module collects outpatient process data and scenario simulation data; the outpatient full-process task scheduling configuration module collects real-time data such as task execution status, visitation queue length, process time, and departmental load; the outpatient intelligent optimization execution module collects data on the execution effect of optimization strategies; the outpatient data archiving module collects dynamic data such as outpatient visitation flow, diagnosis and treatment data, and follow-up visit data in real time. All collected data are preprocessed (noise reduction and standardization) and then synchronously transmitted to the simulation central control module to provide accurate data support for parameter adjustment judgment.

[0066] 1.3) Parameter Adjustment Trigger Judgment: Based on the collected multi-source real-time data and the trigger conditions set according to the characteristics of the outpatient scenario, the simulation central control module automatically determines whether parameter adjustment needs to be initiated. The core trigger conditions are mainly adapted to the outpatient scenario: the change in outpatient traffic is ≥15% (adapting to weekday / weekend, morning / afternoon traffic fluctuations), the average waiting time of the patient queue exceeds the preset threshold (e.g., 20 minutes), the department load deviation is ≥12%, the optimization effect error is ≥7%, and the simulation scenario changes (e.g., general outpatient / specialist outpatient, initial visit / follow-up visit scenario switching). If any of the above conditions are met, a parameter adjustment trigger command is immediately issued to the outpatient dynamic parameter adjustment and optimization module; otherwise, return to step 1.2) and continue to collect real-time data.

[0067] 1.4) Dynamic Parameter Optimization: After receiving the parameter adjustment trigger command, the outpatient dynamic parameter adjustment module calls the improved reinforcement learning parameter adjustment algorithm. Combining the historical parameter adjustment data stored in the outpatient data archiving module, the current real-time operation data, the parameters of the inferred scenario, and the optimization effect feedback data, it focuses on dynamically optimizing and adjusting the outpatient-specific parameters. It prioritizes adjusting core parameters such as the patient priority coefficient, traffic fluctuation adaptation parameters, and follow-up visit reminder timing parameters. At the same time, it fine-tunes auxiliary parameters such as the process timing parameters and task scheduling weights of the inferred model to ensure that the adjusted parameters can accurately adapt to the current outpatient scenario, taking into account both patient efficiency and management convenience.

[0068] 1.5) Parameter Verification and Optimization: The outpatient dynamic parameter adjustment and optimization module synchronously transmits the adjusted parameters to the simulation central control module. Based on the adjusted parameters, the simulation central control module, combined with the high-fidelity process model of the outpatient scenario simulation modeling module, initiates a simulation to verify three core indicators: simulation data accuracy (simulation error ≤ 5%), consultation efficiency (average patient waiting time ≤ 15 minutes), and scenario adaptability (simulation adaptability to different outpatient scenarios ≥ 90%). If all indicators meet the preset requirements, proceed to the next step; if any indicator fails to meet the requirements, return to step 1.4 to readjust the parameters until the preset requirements are met.

[0069] 1.6) Parameter Update Archiving: After the parameters are verified, the outpatient dynamic parameter tuning and optimization module updates the model parameters of the entire engine and archives the adjusted parameters, tuning logs, and verification results to the outpatient data archiving module. At the same time, the parameter update information is fed back to the simulation central control module and all related modules to ensure that all modules use the latest parameters to carry out their work and complete a complete parameter adjustment process.

[0070] 1.7) Iterative execution: If the outpatient simulation and optimization task is not completed (e.g., the outpatient operation simulation for the day is not completed, or the specialist outpatient scenario simulation optimization is not completed), return to step 1.2) to continuously collect real-time data, determine parameter tuning needs, and realize dynamic adaptive iterative adjustment of parameters; if the outpatient simulation and optimization task is completed, the parameter adjustment loop terminates, and all parameter tuning-related data are archived and stored to provide support for subsequent model optimization.

[0071] The following is a detailed description of the morning peak scenario in outpatient services, with the specific steps as follows: 2.1) Engine Startup Self-Check: The simulation central control module starts the outpatient simulation intelligent engine, and each module completes the initialization self-check to ensure that each module operates normally and data transmission is smooth; the outpatient data archiving module loads basic data such as historical outpatient operation data, outpatient department layout data, medical staff configuration information, basic information of diagnosis and treatment equipment, and chronic disease patient records to provide data support for simulation and optimization; the outpatient dynamic parameter adjustment and optimization module loads preset initial parameters and completes parameter initialization.

[0072] 2.2) Outpatient Needs Identification: The outpatient intent recognition module collects outpatient management needs (such as "optimizing the specialist outpatient consultation process, shortening patient waiting time, and improving the follow-up visit reminder mechanism") and consultation-related needs through multiple channels such as the outpatient management backend, medical and nursing workstations, and online registration platforms. Combining outpatient diagnosis and treatment data and consultation scenario information, the module accurately captures the core and priority of needs through an improved deep learning intent recognition algorithm, and transmits the recognition results and needs details to the deduction central control module.

[0073] 2.3) Outpatient Scenario Modeling: The simulation central control module issues modeling instructions to the outpatient scenario simulation modeling module. Combining the outpatient intent recognition results and outpatient treatment guidelines, the outpatient scenario simulation modeling module constructs a high-fidelity outpatient full-process simulation model. It completes the division of all path nodes for outpatient registration, triage and consultation, examination and testing, prescription issuance, payment and medication collection, follow-up reminders, and chronic disease follow-up. It configures the priority rules for consultation and the follow-up reminder mechanism, and simulates scenarios such as outpatient traffic fluctuations and changes in the consultation queue. After the modeling is completed, the process model data is fed back to the simulation central control module and simultaneously archived to the outpatient data archiving module.

[0074] 2.4) Full-process task scheduling: Based on the demand identification results and the simulation model, the simulation central control module issues scheduling instructions to the outpatient full-process task scheduling configuration module. This module rationally allocates task priorities and plans execution paths according to outpatient treatment standards and visit priorities, orderly schedules tasks in each link of the outpatient process, monitors task execution status in real time, collects real-time data such as physical examination traffic, department load, and patient waiting time, and synchronously feeds it back to the simulation central control module to provide data support for subsequent optimization.

[0075] 2.5) Intelligent Optimization Execution: The simulation central control module sends optimization instructions to the outpatient intelligent optimization execution module. Based on the simulation model and real-time operational data, this module generates outpatient-specific optimization tasks, starts the outpatient intelligent optimization algorithm, sets multi-objective optimization weights, and executes the optimization process. It focuses on completing tasks such as outpatient traffic prediction, patient queue optimization, departmental load balancing, and follow-up visit reminder timing optimization. After optimization, it outputs a detailed optimization report, including comparison of indicators before and after optimization, details of optimization strategies, and comparison results of historical solutions. The report is fed back to the simulation central control module and simultaneously archived to the outpatient data archiving module.

[0076] 2.6) Dynamic parameter iteration: The outpatient dynamic parameter tuning and optimization module uses real-time data collected by the outpatient full-process task scheduling and configuration module and optimization effect data fed back by the outpatient intelligent optimization execution module to judge the parameter tuning needs in real time according to the above parameter adjustment method, start the dynamic parameter tuning process, and continuously optimize the model parameters to ensure the accuracy of the simulation results and the adaptation of the optimization strategy to the current outpatient scenario, thereby improving the efficiency and adaptability of outpatient management simulation.

[0077] 2.7) Output of simulation results: The simulation central control module integrates the simulation results of outpatient scenarios, task scheduling data, and intelligent optimization results to generate core simulation data such as outpatient traffic prediction reports, department load distribution tables, task scheduling schemes, and optimization strategy suggestions. Simultaneously, it outputs optimization reports and historical scheme comparison results, which are fed back to outpatient managers and relevant medical staff through outpatient management backend, terminal display screens, and other channels to provide accurate support for decision-making.

[0078] 2.8) Implementation of optimization strategies: Based on the simulation data and optimization report, and in combination with the actual treatment situation in the outpatient department, the outpatient managers implement optimization strategies, including optimizing the treatment process, adjusting departmental load, improving the follow-up visit reminder mechanism, and allocating outpatient resources, so as to improve the efficiency of outpatient treatment and management effectiveness and meet the needs of patients for convenient medical treatment.

[0079] 2.9) Termination of simulation process: If the daily outpatient operation simulation and optimization task is completed, the simulation central control module shuts down the engine and archives all relevant data of this simulation (simulation data, optimization report, parameter data, visit data, etc.) to the outpatient data archiving module to complete this simulation process; if it is necessary to continue to carry out simulations (such as specialist outpatient scenario simulation, chronic disease management simulation, etc.), return to step 2.2) to continue to execute the simulation and optimization process.

[0080] The following is a detailed description of the emergency service scenario, with specific steps as follows: 1.1) Initialization Configuration: The simulation central control module starts the emergency simulation intelligent agent engine. Each module completes the initialization self-check. The emergency dynamic parameter tuning engine module loads the preset initial parameters. The initial parameters are preset based on the routine operation scenario of the emergency department, clinical diagnosis and treatment guidelines and historical data. They mainly cover three types of core parameters: simulation modeling parameters (critical care priority judgment threshold, service time distribution of each process node, emergency scenario simulation parameters, etc.), task optimization weight parameters (critical care emergency time weight, equipment utilization weight, patient waiting time weight, etc.), and queuing scheduling parameters (priority coefficient, green channel triggering parameters, etc.). All initial parameters are synchronously archived to the emergency data archiving module to ensure parameter traceability.

[0081] 1.2) Multi-source data acquisition: The emergency intention perception module collects data on emergency management and operation needs and clinical treatment-related needs; the emergency scenario simulation modeling module collects data on emergency clinical processes and scenario simulation data; the emergency full-process scheduling and control module collects real-time data such as task execution status, departmental load, patient waiting time, and medical staff work status; the emergency intelligent optimization execution module collects data on the execution effect of optimization strategies; the emergency data archiving module collects dynamic data such as physical examination traffic, emergency equipment operation status, and treatment time for critically ill patients in real time. All collected data are preprocessed and synchronously transmitted to the simulation central control module to provide data support for parameter adjustment judgment.

[0082] 1.3) Parameter Adjustment Trigger Judgment: Based on the collected multi-source real-time data and the trigger conditions set according to the characteristics of the emergency scenario, the simulation central control module automatically determines whether parameter adjustment needs to be initiated. The core trigger conditions include: emergency traffic change ≥ 20% (adapting to scenarios of sudden increase / decrease in emergency traffic), departmental load deviation ≥ 15% (adapting to scenarios of departmental load imbalance), critical care emergency treatment time exceeding the preset threshold (e.g., 30 minutes, to ensure the efficiency of critical care treatment), optimization effect error ≥ 8% (to ensure the effectiveness of optimization strategies), and simulation scenario switching (e.g., off-peak / peak switching, routine emergency / sudden emergency switching). If any of the above conditions are met, a parameter adjustment trigger command is immediately issued to the emergency dynamic parameter adjustment engine module; otherwise, return to step 1.2) and continue to collect real-time data.

[0083] 1.4) Dynamic Parameter Adjustment: After receiving the parameter adjustment trigger command, the emergency dynamic parameter tuning engine module calls the improved reinforcement learning parameter tuning algorithm. Combining the historical parameter adjustment data stored in the emergency data archiving module, the current real-time operation data, the parameters of the inferred scenario, and the optimization effect feedback data, it focuses on dynamically optimizing and adjusting the emergency-specific parameters. It prioritizes adjusting core parameters such as the critical care priority judgment threshold, emergency equipment scheduling parameters, and peak period optimization weights. At the same time, it fine-tunes auxiliary parameters such as the node service time distribution and queuing scheduling coefficient of the inferred model to ensure that the adjusted parameters can accurately adapt to the current emergency scenario and take into account both the priority of critical care and the overall operational efficiency.

[0084] 1.5) Parameter Verification and Optimization: The emergency dynamic parameter tuning engine module synchronously transmits the adjusted parameters to the simulation central control module. Based on the adjusted parameters, the simulation central control module, combined with the high-fidelity process model of the emergency scenario simulation modeling module, initiates a simulation to verify four core indicators: simulation data accuracy (simulation error ≤ 5%), critical care treatment efficiency (critical care emergency treatment time ≤ 25 minutes), resource utilization efficiency (emergency equipment utilization rate ≥ 85%), and patient experience (average waiting time for ordinary emergency patients ≤ 15 minutes). If all indicators meet the preset requirements, proceed to the next step; if any indicator fails to meet the requirements, return to step 1.4), readjust the parameters, and continue until the preset requirements are met.

[0085] 1.6) Parameter Update Archiving: After the parameters are verified, the emergency dynamic parameter tuning engine module updates the model parameters of the entire engine and archives the adjusted parameters, tuning logs, and verification results to the emergency data archiving module. At the same time, the parameter update information is fed back to the simulation central control module and all related modules to ensure that all modules use the latest parameters to carry out their work and complete a complete parameter adjustment process.

[0086] 1.7) Iterative execution: If the emergency simulation and optimization task is not completed (e.g., the emergency operation simulation for the day is not completed, or the emergency scenario simulation optimization is not completed), return to step 1.2) to continuously collect real-time data, determine parameter tuning requirements, and realize dynamic adaptive iterative adjustment of parameters; if the emergency simulation and optimization task is completed, the parameter adjustment loop terminates, and all parameter tuning-related data is archived and stored to provide support for subsequent model optimization.

[0087] The following is a detailed description of peak-hour scenarios in emergency services, with the specific steps as follows: 2.1) Engine Startup Self-Check: The simulation central control module starts the emergency simulation intelligent agent engine, and each module completes the initialization self-check to ensure that each module operates normally and data transmission is smooth; the emergency data archiving module loads basic data such as historical emergency operation data, emergency department layout data, basic information of emergency equipment, and medical staff configuration information to provide data support for simulation and optimization; the emergency dynamic parameter tuning engine module loads preset initial parameters and completes parameter initialization.

[0088] 2.2) Emergency Needs Perception: The Emergency Intent Perception module collects emergency management and operational needs (such as "to cope with the evening emergency peak, optimize the load of internal medicine and surgical departments, adjust the allocation of medical staff, and shorten the time for emergency treatment of critically ill patients") and clinical treatment needs (such as "to open a green channel for stroke patients and prioritize the arrangement of rescue equipment") through multiple channels such as the emergency management backend and doctor workstations. Through the improved BERT intent recognition algorithm, it accurately captures the core and priority of the needs and transmits the recognition results and needs details to the deduction central control module.

[0089] 2.3) Emergency Scenario Modeling: The simulation central control module issues modeling instructions to the emergency scenario simulation modeling module. Combining the emergency intent perception results and clinical diagnosis and treatment guidelines, the emergency scenario simulation modeling module constructs a high-fidelity emergency full-process simulation model. It completes the division of all path nodes for patient arrival, triage, priority determination, diagnostic examination, critical care rescue, inpatient transfer, and discharge (outpatient visit). It configures a critical care priority determination mechanism and green channel triggering rules, sets multiple queuing scheduling strategies and node flow rules, and simulates sudden situations such as traffic fluctuations, sudden changes in condition, and equipment failures in the emergency scenario. After the modeling is completed, the process model data is fed back to the simulation central control module and simultaneously archived to the emergency data archiving module.

[0090] 2.4) Full-process task scheduling: Based on the demand perception results and the simulation model, the simulation central control module issues scheduling instructions to the emergency full-process scheduling and control module. This module schedules emergency full-process tasks in an orderly manner according to the preset queuing scheduling strategy and node flow rules, giving priority to the treatment of critically ill patients, monitoring the task execution status in real time, collecting real-time data such as physical examination traffic, department load, medical staff work status, and patient waiting time, and synchronously feeding it back to the simulation central control module to provide data support for subsequent optimization.

[0091] 2.5) Intelligent Optimization Execution: The simulation central control module sends optimization instructions to the emergency intelligent optimization execution module. Based on the simulation model and real-time operational data, this module generates multi-dimensional emergency optimization tasks, activates emergency-specific optimization algorithms, sets multi-objective optimization weights, and executes the optimization process. It focuses on completing optimization tasks such as emergency traffic prediction, critical care admission path planning, medical staff scheduling adjustment, and emergency equipment allocation. At the same time, it activates the event traceability function to fully record the data of the entire patient treatment process. After optimization, it outputs a detailed optimization report, including comparison of indicators before and after optimization, details of optimization strategies, and comparison results of historical plans, which is fed back to the simulation central control module and simultaneously archived to the emergency data archiving module.

[0092] 2.6) Dynamic parameter iteration: The emergency dynamic parameter tuning engine module judges the parameter tuning needs in real time according to the real-time data collected by the emergency full-process scheduling and control module and the optimization effect data fed back by the emergency intelligent optimization execution module, and starts the dynamic parameter tuning process to continuously optimize the model parameters, ensuring that the simulation results are accurate, the optimization strategy is adapted to the current emergency scenario, and the efficiency of emergency treatment is always at its best.

[0093] 2.7) Output of simulation results: The simulation central control module integrates the simulation results of emergency scenarios, task scheduling data, and intelligent optimization results to generate core simulation data such as emergency traffic prediction reports, departmental load distribution tables, medical staff allocation plans, and emergency equipment scheduling suggestions. It also outputs optimization reports and historical plan comparison results simultaneously. These are fed back to emergency managers and relevant medical staff through channels such as the emergency management backend and terminal displays, providing accurate support for decision-making.

[0094] 2.8) Implementation of optimization strategies: Based on the simulation data and optimization report, and in combination with the actual clinical situation, emergency department managers implement optimization strategies, including adjusting departmental load distribution, optimizing medical staff scheduling, allocating emergency equipment, and implementing the optimal admission pathway for critically ill patients, to prioritize the treatment efficiency of critically ill patients and improve the overall operational efficiency of the emergency department.

[0095] 2.9) Termination of simulation process: If the emergency operation simulation and optimization task is completed on the same day, the simulation central control module shuts down the engine and archives all relevant data of this simulation (simulation data, optimization report, parameter data, event backtracking data, etc.) to the emergency data archiving module to complete this simulation process; if simulation needs to be carried out continuously (such as sudden emergency traffic peak, special scenario simulation), return to step 2.2) to continue to execute the simulation and optimization process.

[0096] The following is a detailed description of the inpatient service scenario, with specific steps as follows: 1.1) Initialization Configuration: The central control module of the simulation starts the intelligent engine for inpatient management simulation. Each module completes the initialization self-check. The inpatient dynamic parameter tuning and optimization module loads the preset initial parameters. The initial parameters are preset based on the routine operation scenario of inpatient care, treatment guidelines and historical inpatient data. They mainly cover three types of core parameters: inpatient simulation modeling parameters (disease classification threshold, bed allocation parameters, treatment time sequence parameters, etc.), task optimization weight parameters (bed utilization rate weight, treatment efficiency weight, patient recovery effect weight, etc.), and parameter tuning and adaptation parameters (disease change adaptation coefficient, bed fluctuation adaptation parameters, etc.). All initial parameters are synchronously archived to the inpatient data archiving module to ensure that the parameters are traceable and reviewable.

[0097] 1.2) Multi-source data acquisition: The inpatient intent accurate identification module collects inpatient management needs and inpatient scenario information; the inpatient scenario simulation modeling module collects inpatient process data and scenario simulation data; the inpatient full-process task scheduling and configuration module collects real-time data such as task execution status, bed occupancy, treatment progress, and follow-up completion rate; the inpatient intelligent optimization execution module collects data on the execution effect of optimization strategies; the inpatient data archiving module collects dynamic data such as inpatient condition data, bed resource data, treatment data, and discharge follow-up data in real time. All collected data are preprocessed (noise reduction and standardization) and then synchronously transmitted to the simulation central control module to provide accurate data support for parameter adjustment judgment.

[0098] 1.3) Parameter Adjustment Trigger Judgment: Based on the collected multi-source real-time data and the trigger conditions set according to the characteristics of the inpatient scenario, the simulation central control module automatically determines whether parameter adjustment needs to be initiated. The core trigger conditions are different from those for outpatient, emergency and health management, and are mainly adapted to the inpatient scenario: the change in bed occupancy rate is ≥10% (adapting to bed resource fluctuations), the patient's condition grade changes (such as mild cases turning into severe cases, severe cases turning into the recovery period), the diagnosis and treatment efficiency error is ≥6%, the optimization effect error is ≥7%, and the simulation scenario changes (such as the change between routine inpatient / emergency inpatient and severe / recovery inpatient scenarios). If any of the above conditions are met, a parameter adjustment trigger command is immediately issued to the inpatient dynamic parameter adjustment and optimization module; otherwise, return to step 1.2) and continue to collect real-time data.

[0099] 1.4) Dynamic Parameter Optimization: After receiving the parameter adjustment trigger command, the inpatient dynamic parameter adjustment module calls the improved reinforcement learning parameter adjustment algorithm. Combining the historical parameter adjustment data stored in the inpatient data archiving module, the current real-time operation data, the parameters of the inferential scenario, and the optimization effect feedback data, it focuses on dynamically optimizing and adjusting the inpatient-specific parameters. It prioritizes adjusting core parameters such as bed allocation parameters, patient condition matching parameters, diagnosis and treatment time sequence parameters, and follow-up optimization parameters. At the same time, it fine-tunes auxiliary parameters such as the task priority coefficient and optimization weight of the inferential modeling to ensure that the adjusted parameters can accurately adapt to the current inpatient scenario and take into account the diagnosis and treatment efficiency, rehabilitation effect, and bed utilization rate.

[0100] 1.5) Parameter Verification and Optimization: The inpatient dynamic parameter adjustment and optimization module synchronously transmits the adjusted parameters to the simulation central control module. Based on the adjusted parameters, the simulation central control module, combined with the high-fidelity process model of the inpatient scenario simulation modeling module, initiates a simulation to verify three core indicators: simulation data accuracy (simulation error ≤ 5%), inpatient management efficiency (bed utilization rate ≥ 85%, treatment delay rate ≤ 3%), and scenario adaptability (simulation adaptability to different inpatient scenarios ≥ 90%). If all indicators meet the preset requirements, proceed to the next step; if any indicator fails to meet the requirements, return to step 1.4) to readjust the parameters until the preset requirements are met.

[0101] 1.6) Parameter Update Archiving: After the parameters are verified, the inpatient dynamic parameter tuning and optimization module updates the model parameters of the entire engine and archives the adjusted parameters, tuning logs, and verification results to the inpatient data archiving module. At the same time, the parameter update information is fed back to the simulation central control module and all related modules to ensure that all modules use the latest parameters to carry out their work and complete a complete parameter adjustment process.

[0102] 1.7) Iterative execution: If the inpatient simulation and optimization task is not completed (e.g., the inpatient operation simulation for the current month is not completed, or the critical care patient diagnosis and treatment simulation is not completed), return to step 1.2) to continuously collect real-time data, determine parameter tuning needs, and realize dynamic adaptive iterative adjustment of parameters; if the inpatient simulation and optimization task is completed, the parameter adjustment loop terminates, and all parameter tuning-related data are archived and stored to provide support for subsequent model optimization.

[0103] The following is a detailed description of peak-hour scenarios in inpatient services, with the specific steps as follows: 2.1) Engine Startup Self-Check: The simulation central control module starts the inpatient management simulation intelligent engine, and each module completes the initialization self-check to ensure that each module operates normally and data transmission is smooth; the inpatient data archiving module loads basic data such as historical inpatient operation data, inpatient department layout data, bed resource information, medical staff configuration information, and patient medical records to provide data support for simulation and optimization; the inpatient dynamic parameter tuning and optimization module loads preset initial parameters and completes parameter initialization.

[0104] 2.2) Hospitalization Needs Identification: The accurate hospitalization intent identification module collects hospitalization management needs (such as "optimizing the diagnosis and treatment process for critically ill internal medicine patients, improving bed utilization, improving the discharge follow-up mechanism, and shortening the patient's hospitalization period") and diagnosis and treatment related needs through multiple channels such as the hospitalization management backend, medical and nursing workstations, and nursing terminals. Combining hospitalization diagnosis and treatment data and hospitalization scenario information, the module accurately captures the core and priority of needs through an improved deep learning intent identification algorithm, and transmits the identification results and needs details to the deduction central control module.

[0105] 2.3) Inpatient Scenario Modeling: The central control module of the simulation sends modeling instructions to the inpatient scenario simulation modeling module. Combining the results of inpatient intent recognition and inpatient treatment guidelines, the inpatient scenario simulation modeling module constructs a high-fidelity inpatient full-process simulation model. It completes the division of all path nodes for inpatient admission assessment, treatment implementation, process control, discharge planning, and discharge follow-up. It configures the rules for disease grading, bed allocation, discharge assessment standards, and follow-up timing, and simulates scenarios such as dynamic changes in patient condition, fluctuations in bed resources, and multi-departmental collaborative diagnosis and treatment. After the modeling is completed, the process model data is fed back to the central control module of the simulation and simultaneously archived to the inpatient data archiving module.

[0106] 2.4) Full-process task scheduling: Based on the demand identification results and the simulation model, the simulation central control module issues scheduling instructions to the inpatient full-process task scheduling configuration module. This module, in accordance with the inpatient treatment guidelines, reasonably allocates task priorities, plans execution paths, and orderly schedules tasks in each stage of inpatient care. It monitors the task execution status in real time, collects real-time data such as bed occupancy, treatment progress, and follow-up completion rate, and synchronously feeds it back to the simulation central control module to provide data support for subsequent optimization.

[0107] 2.5) Intelligent Optimization Execution: The simulation central control module issues optimization instructions to the inpatient intelligent optimization execution module. Based on the simulation model and real-time operational data, this module generates inpatient-specific optimization tasks, starts the inpatient intelligent optimization algorithm, sets multi-objective optimization weights, executes the optimization process, and focuses on completing tasks such as inpatient bed allocation, multi-department collaborative diagnosis and treatment path planning, diagnosis and treatment progress optimization, and discharge follow-up sequence optimization. After optimization is completed, a detailed optimization report is output, including comparison of indicators before and after optimization, details of optimization strategies, and comparison results of historical plans, which is fed back to the simulation central control module and simultaneously archived to the inpatient data archiving module.

[0108] 2.6) Dynamic parameter iteration: The inpatient dynamic parameter tuning and optimization module uses real-time data collected by the inpatient full-process task scheduling and configuration module and optimization effect data fed back by the inpatient intelligent optimization execution module to judge the parameter tuning needs in real time according to the above parameter adjustment method, start the dynamic parameter tuning process, and continuously optimize the model parameters to ensure the accuracy of the simulation results and the adaptation of the optimization strategy to the current inpatient scenario, thereby improving the efficiency and adaptability of inpatient management simulation.

[0109] 2.7) Output of simulation results: The simulation central control module integrates the simulation results of inpatient scenarios, task scheduling data, and intelligent optimization results to generate core simulation data such as inpatient bed prediction reports, departmental treatment load distribution tables, bed allocation plans, and optimization strategy suggestions. Simultaneously, it outputs optimization reports and historical plan comparison results, which are fed back to inpatient managers and relevant medical staff through channels such as the inpatient management backend and terminal display screens, providing accurate support for decision-making.

[0110] 2.8) Implementation of optimization strategies: Based on the simulation data and optimization report, and in combination with the actual inpatient treatment situation, the inpatient manager implements optimization strategies, including optimizing bed allocation, adjusting treatment processes, improving the discharge follow-up mechanism, and allocating medical and nursing staff, so as to improve the efficiency of inpatient management, bed utilization rate and patient recovery effect, and meet the needs of diverse inpatient management scenarios.

[0111] 2.9) Termination of simulation process: If the monthly inpatient operation simulation and optimization task is completed, the simulation central control module shuts down the engine and archives all relevant data of this simulation (simulation data, optimization report, parameter data, inpatient diagnosis and treatment data, etc.) to the inpatient data archiving module to complete this simulation process; if it is necessary to continue to carry out simulations (such as critical patient diagnosis and treatment simulation, bed resource optimization simulation, etc.), return to step 2.2) to continue to execute the simulation and optimization process.

[0112] See Figure 10 As shown in the figure, this application discloses a medical service task scheduling device, including: The intent recognition module 11 is used to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients through the BERT intent recognition algorithm to obtain the core intent of medical service scheduling. The model building module 12 is used to construct a medical service process model based on the core intent of the medical service scheduling, and configure the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively. The model simulation module 13 is used to input the historical operation data and scenario data of the medical service center into the scheduling simulation model and perform simulation to obtain the scheduling results; The model optimization module 14 is used to adaptively adjust the parameters of the scheduling simulation model based on the scheduling result and through a reinforcement learning algorithm when the historical operation data meets the preset parameter adjustment triggering conditions, so as to obtain a new scheduling simulation model, and then jump back to the step of inputting the historical operation data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so as to determine the obtained scheduling simulation model as the target simulation model. The medical service scheduling module 15 is used to input the current operating data of the medical service center into the target simulation model in order to output a scheduling scheme.

[0113] Furthermore, embodiments of this application also provide an electronic device. Figure 11This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0114] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Specifically, it may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the medical service task scheduling method performed by the electronic device disclosed in any of the foregoing embodiments.

[0115] In this embodiment, the power supply 23 is used to provide operating voltage for various hardware devices on the electronic device; the communication interface 24 can create a data transmission channel between the electronic device and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0116] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0117] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0118] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. The operating system can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the medical service task scheduling method executed by the electronic device as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0119] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned disclosed medical service task scheduling method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0120] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0121] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. The software module may be located in random access memory (RAM), memory, read-only memory (ROM), electrically programmable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), register, hard disk, removable disk, CD-ROM (Compact Disc Read-Only Memory), or any other form of storage medium known in the art.

[0122] Finally, 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 a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0123] The above provides a detailed description of a medical service task scheduling method, apparatus, equipment, and medium provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only intended to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for scheduling medical service tasks, characterized in that, include: The BERT intent recognition algorithm is used to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent of medical service scheduling. A medical service process model is constructed based on the core intent of the medical service scheduling, and the node flow rules and multi-mode queuing scheduling strategy of the medical service process model are configured to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively. The historical operational data and scenario data of the medical service center are input into the scheduling simulation model, and the simulation is executed to obtain the scheduling results. When the historical operational data meets the preset parameter adjustment triggering conditions, the parameters of the scheduling simulation model are adaptively adjusted based on the scheduling results and through a reinforcement learning algorithm to obtain a new scheduling simulation model. Then, the process jumps back to the step of inputting the historical operational data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so that the obtained scheduling simulation model is determined as the target simulation model. Input the current operational data of the medical service center into the target simulation model to output a scheduling scheme.

2. The medical service task scheduling method according to claim 1, characterized in that, The process involves using the BERT intent recognition algorithm to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients to obtain the core intent for medical service scheduling, including: Collect information on the operational needs of medical service centers and the medical service needs of medical service recipients; The operational requirements and medical service requirements are preprocessed and feature extracted to obtain scheduling intention features; The BERT deep learning model is used to match keywords of medical service scenarios with the scheduling intent features to obtain the core intent of medical service scheduling. The core intent of medical service scheduling includes any one or more of the following: optimization of clinic / bed resources, adjustment of medical service schedule, traffic prediction, queuing time control, priority scheduling of critically ill patients, allocation of medical staff, equipment scheduling, and follow-up / revisit planning.

3. The medical service task scheduling method according to claim 1, characterized in that, The step of constructing a medical service process model based on the core intent of the medical service scheduling, and configuring the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model includes: Based on the core intent of the medical service scheduling, each medical service process node is constructed, and the timing and logical relationship of each medical service process node are defined to obtain the medical service process model of each medical service process node. The node jump and multi-path branch rules of the medical service process model are set, and a multi-mode queuing scheduling strategy is configured based on the first-in-first-out scheduling strategy and the priority scheduling strategy to obtain a deducible scheduling simulation model.

4. The medical service task scheduling method according to claim 1, characterized in that, The step of adaptively adjusting the parameters of the scheduling simulation model based on the scheduling results and using a reinforcement learning algorithm to obtain a new scheduling simulation model includes: Based on the scheduling results and the current medical service scenario, and through reinforcement learning algorithms, the node service time distribution, task optimization weight, and queuing strategy coefficient in the scheduling simulation model are dynamically optimized to generate updated model parameters that are adapted to the current operating status of the medical service center. The model is reconstructed based on the updated model parameters to obtain a new scheduling simulation model.

5. The medical service task scheduling method according to claim 1, characterized in that, The historical operational data includes the actual change in medical service traffic, the actual utilization rate deviation of consultation rooms / beds, the actual queuing time, and the actual medical service scenarios. The preset parameter adjustment trigger condition is at least one of the following conditions: The actual change in the medical service flow is greater than the first preset threshold. The actual utilization rate of the consultation room / bed deviates from the second preset threshold. The actual queuing time is greater than the third preset threshold; The actual medical service scenario represents the state switching of the medical service scenario.

6. The medical service task scheduling method according to claim 5, characterized in that, The scheduling results include the predicted change in medical service traffic, the predicted utilization rate deviation of clinics / beds, the predicted queuing time, and the predicted medical service scenario; the adaptive adjustment of the parameters of the scheduling simulation model based on the scheduling results and using a reinforcement learning algorithm includes: Based on the target error between the scheduling result and the historical operation data, the parameters of the scheduling simulation model are adaptively adjusted using a reinforcement learning algorithm. The target error includes a first error between the predicted change in medical service traffic and the actual change in medical service traffic, a second error between the predicted utilization rate of consultation rooms / beds and the actual utilization rate of consultation rooms / beds, a third error between the predicted queuing time and the actual queuing time, and a fourth error between the predicted medical service scenario and the actual medical service scenario.

7. The medical service task scheduling method according to any one of claims 1 to 6, characterized in that, Also includes: The core intent of the medical service scheduling, the model parameters of the scheduling simulation model, the operational data of the medical service center, the scheduling results, and the scheduling scheme are encrypted to obtain encrypted data; wherein, the scheduling scheme includes any one or more of the following: consultation room / bed allocation scheme, patient examination scheduling, medical staff shift scheduling, equipment allocation scheme, triage path planning, follow-up reminder, and follow-up planning. The encrypted data is saved to a distributed database.

8. A medical service task scheduling device, characterized in that, include: The intent recognition module is used to analyze the operational needs of the medical service center and the medical service needs of the medical service recipients through the BERT intent recognition algorithm to obtain the core intent of medical service scheduling. The model building module is used to construct a medical service process model based on the core intent of the medical service scheduling, and configure the node flow rules and multi-mode queuing scheduling strategy of the medical service process model to obtain a deducible scheduling simulation model; wherein, the process nodes of the medical service process model are the process nodes corresponding to physical examination service tasks, emergency service tasks, outpatient service tasks and inpatient service tasks respectively. The model simulation module is used to input the historical operation data and scenario data of the medical service center into the scheduling simulation model and perform simulation to obtain the scheduling results. The model optimization module is used to adaptively adjust the parameters of the scheduling simulation model based on the scheduling result and through a reinforcement learning algorithm when the historical operation data meets the preset parameter adjustment triggering conditions, so as to obtain a new scheduling simulation model. Then, it jumps back to the step of inputting the historical operation data and scenario data of the medical service center into the scheduling simulation model until a scheduling result that meets the preset deduction and verification conditions is obtained, so as to determine the obtained scheduling simulation model as the target simulation model. The medical service scheduling module is used to input the current operating data of the medical service center into the target simulation model in order to output a scheduling plan.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the medical service task scheduling method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the medical service task scheduling method as described in any one of claims 1 to 7.

Images