Active nursing service scheduling system and method based on multi-modal intention prediction

Abstract

The invention discloses an active nursing service scheduling system and method based on multi-modal intention prediction, and belongs to the technical field of smart pension and artificial intelligence scheduling. In order to solve the problems of delay, low efficiency and non-uniform resource allocation in the link from'intention sensing 'to'service execution' in the existing nursing mode, the system comprises an intention signal receiving unit used for receiving structured intention prediction generated by an external system; a static rule engine and a dynamic resource optimization module are arranged in the service scheduling decision core unit, and intelligent decision making is carried out according to the real-time resource state; the resource state management unit monitors and updates a state database of nursing manpower and equipment in real time; and the task distribution and feedback unit is responsible for issuing an instruction and receiving a completion signal to form a closed loop. Through cooperative work of the units, full-process automation from intention receiving to task distribution is achieved, cost, efficiency, priority and personalized preference can be comprehensively considered, optimal scheduling is conducted on nursing resources, meanwhile, the response speed and resource utilization efficiency of nursing services are remarkably improved, and the workload of nursing staff is reduced.

Description

Technical Field

[0001] This invention relates to artificial intelligence technology, Internet of Things technology, automated scheduling technology, and the fields of smart health and elderly care services. More specifically, this invention relates to a system and method that, upon receiving a prediction of a service recipient's intentions, can automatically and intelligently plan and schedule nursing resources to perform corresponding services. Background Technology

[0002] Currently, the global population is aging rapidly, creating unprecedented demands for efficient, precise, and humane elderly care services. For disabled elderly people who have lost their ability to speak or move due to stroke, neurodegenerative diseases, or other reasons, accurately expressing their needs is crucial for maintaining their quality of life and dignity. Existing care models are mostly reactive, relying on the elderly using physical call buttons or on caregivers making regular rounds to identify needs. This model has a significant lag, failing to promptly meet the elderly's sudden or potential needs, and also increasing the workload of caregivers.

[0003] To address the challenge of expressing needs, several intent recognition technologies based on multimodal information have emerged in recent years. For example, by analyzing nonverbal cues such as eye movements, gaze patterns, physiological signals, and environmental information, artificial intelligence models are used to predict the elderly's potential intentions. These technologies have made significant progress in "perceiving" and "understanding" the needs of the elderly, generating relatively accurate intent prediction signals.

[0004] However, existing technologies exhibit significant gaps and technological deficiencies in the transition from "understanding intent" to "serve execution." Even if a system successfully predicts an elderly person's intention to "drink water," translating this information into a specific and efficient care task and allocating it to the most appropriate resources still relies on human judgment and manual scheduling. Administrators of nursing homes or retirement communities need to make ad-hoc assignments based on each caregiver's current location, work status, and workload. This manual scheduling method has the following drawbacks:

[0005] Response delay: There is a human decision-making delay between the generation of intent and the execution of service, which reduces the timeliness of service and may pose risks, especially for urgent needs.

[0006] Uneven resource allocation: Manual scheduling makes it difficult to grasp the precise status of all resources in real time, which can easily lead to some nursing staff being overloaded while other resources are idle, resulting in a waste of human resources and occupational burnout among nursing staff.

[0007] Inefficiency: The lack of comprehensive consideration of factors such as task priority, resource execution efficiency, and service costs often results in scheduling decisions that are not optimal, leading to low overall nursing service efficiency.

[0008] Lack of closed-loop management: The execution status and results of tasks are difficult to be automatically tracked and recorded by the system, which is not conducive to the formation of a data-driven service quality monitoring and continuous improvement mechanism.

[0009] Therefore, given the existing technological capability to predict intentions, the key technical challenge in the field of smart elderly care is how to build a system that can automatically receive intention signals and perform intelligent, optimized, and proactive service scheduling to achieve seamless integration from "perception" to "action." Summary of the Invention

[0010] This invention aims to address the problems mentioned in the background section, namely, bridging the technological gap between intent recognition and service execution. Specifically, this invention aims to provide a proactive nursing service scheduling system and method. This system can automatically receive intent prediction signals and, based on real-time status perception of nursing resources and intelligent decision-making algorithms, achieve efficient, accurate, and automated scheduling of human or equipment resources, thereby significantly improving the response speed, resource utilization, and overall service quality of nursing services. The specific technical solution is as follows:

[0011] A proactive nursing service scheduling system based on multimodal intent prediction includes: a. An intent signal receiving unit, configured to receive a structured intent prediction signal generated by an external intent recognition system, the signal containing at least a unique identifier of the service object, one or more candidate intents and their confidence levels; b. A service scheduling decision core unit, connected to the intent signal receiving unit, configured to parse the intent prediction signal and generate the optimal service execution instruction based on preset scheduling logic and real-time resource status information; c. A resource status management unit, connected to the service scheduling decision core unit, configured to monitor and dynamically update a status database containing the availability, location, and workload of multiple types of nursing resources in real time; d. A task dispatch and feedback unit, connected to the service scheduling decision core unit and the resource status management unit, is configured to issue the service execution instruction to the designated nursing resources, and receive a task completion feedback signal after the service is completed, so as to trigger an update of the status database in the resource status management unit.

[0012] Preferably, the intent prediction signal also includes the urgency level of the intent and the geospatial information of where the intent occurred.

[0013] Preferably, the service scheduling decision core unit includes a static service rule engine that stores a series of intent-service mapping rules to handle high-confidence and routine care needs, enabling rapid response scheduling.

[0014] Preferably, the service scheduling decision core unit also includes a dynamic resource optimization module, which is activated when processing complex or ambiguous intentions and is configured to comprehensively evaluate at least one of the following pieces of information to make a scheduling decision: (1) Estimated time and energy costs for each available nursing resource to perform a specific task; (2) Historical preference data or personalized care plans of service recipients; (3) The priority queue of all pending tasks.

[0015] Preferably, the dynamic resource optimization module employs a weighted evaluation algorithm, the mathematical model of which is:

[0016]

[0017] in, For resources Regarding intention The final scheduling score, For cost function, For users The preference function, To execute the efficiency function, These are configurable weighting coefficients.

[0018] Preferably, the human resource status in the resource status management unit includes at least three types: "idle", "on a task", and "resting", and the equipment resource status includes at least four types: "available", "running", "charging", and "faulty".

[0019] Preferably, the task dispatch and feedback unit sends service execution instructions in the form of structured data to the control system of the mobile smart terminal or nursing equipment held by the nursing staff via a wireless communication network.

[0020] Preferably, the method includes the following steps: (1) Receiving steps: Receive a structured intent prediction signal containing the service object identifier and potential intent through the system's communication interface; (2) Decision-making steps: The service scheduling decision core unit is invoked. First, based on the confidence and clarity of the intent, the static service rule engine or the dynamic resource optimization module is selected for analysis. In this step, the system queries a real-time updated resource status database to obtain the status of all available nursing manpower and equipment. (3) Generation steps: Based on the decision analysis results, generate a service execution instruction that includes task type, target resources, and execution parameters; (4) Dispatch step: Send the service execution instruction to the selected nursing resource terminal; (5) Closed-loop feedback step: After receiving the signal that the task is completed, update the status of the corresponding resource in the resource status database to complete a closed-loop scheduling of proactive service.

[0021] Preferably, in the decision-making step, if the dynamic resource optimization module is activated, the module will execute an optimization algorithm whose goal is to minimize the total scheduling cost of nursing resources while meeting service quality requirements.

[0022] Preferably, the method also includes a logging step to structure and store each intent signal, decision-making process, execution instruction, and task feedback result, providing data support for subsequent service model analysis and system optimization.

[0023] The beneficial effects of the proactive nursing service scheduling system and method based on multimodal intent prediction provided by this invention include:

[0024] 1. Proactive service and rapid response: This system transforms passive response into proactive service. Once an intent is identified, the scheduling process is automatically triggered without manual intervention. In particular, the use of a static rule engine to handle routine requests significantly reduces the delay from request generation to service execution, improving service immediacy.

[0025] 2. The system achieves intelligent and optimal allocation of nursing resources: Through the dynamic resource optimization module, the system can make intelligent decisions based on real-time data in multiple dimensions and with multiple objectives, taking into account cost, efficiency, priority and personalized preferences, so as to achieve dynamic optimal allocation of nursing resources and avoid the problems of resource waste and uneven distribution.

[0026] 3. Improved efficiency and transparency in nursing management: Real-time monitoring of resource status and closed-loop management of task processes enable managers to clearly and transparently grasp the operational status of the entire nursing system. All scheduling and execution data are recorded, providing objective data for subsequent service quality assessments, personnel performance evaluations, and process optimization.

[0027] 4. Reduced workload and occupational stress for nursing staff: Automated task assignment replaces traditional manual assignment and verbal communication. Nursing staff only need to execute tasks according to clear instructions received on their mobile devices and provide simple status feedback. This reduces communication costs and decision-making burden, allowing them to focus more on core nursing work, thereby reducing burnout.

[0028] 5. Enhanced system scalability and compatibility: Through standardized interface design, the system can be easily connected to intent recognition systems from different manufacturers and different types of intelligent nursing devices, exhibiting good compatibility and scalability, making it easy to deploy and upgrade in elderly care institutions of different sizes and types. Attached Figure Description

[0029] Figure 1 This is a block diagram of the overall architecture of the proactive nursing service scheduling system described in this invention;

[0030] Figure 2 This is an internal logical structure diagram of the service scheduling decision core unit described in this invention;

[0031] Figure 3 This is a flowchart of the proactive nursing service scheduling method described in this invention. Detailed Implementation

[0032] This invention proposes a proactive nursing service scheduling system and method based on multimodal intent prediction. Its core is a proactive nursing service scheduling system integrating real-time perception, intelligent decision-making, and closed-loop control. This system acts as a central platform, connecting upstream to various intent recognition systems and downstream to various nursing resource execution terminals. The invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0033] A proactive nursing service scheduling system based on multimodal intent prediction includes: a. An intent signal receiving unit, configured to receive structured intent prediction signals generated by an external intent recognition system. The signal contains at least a unique identifier of the service object, one or more candidate intents and their confidence levels, and the intent prediction signal also includes the urgency level of the intent and the geospatial information of where the intent occurred. Specifically, this unit serves as the system's standard input interface, responsible for receiving structured intent prediction signals generated by the front-end multimodal intent recognition system. To ensure the accuracy and completeness of information transmission, the signal is designed in a standardized data format, typically JSON or a similar format. Its content includes not only the core unique identifier of the service object and the most likely intent description, but also a series of important metadata. This metadata includes: a confidence score for the intent prediction, used to measure the reliability of the prediction; a preset urgency level for subsequent task prioritization; and geospatial information of when the intent occurred, providing a basis for route planning and resource scheduling. This structured signal design provides rich and standardized input for subsequent automated decision-making.

[0034] b. A service scheduling decision core unit, connected to the intent signal receiving unit, is configured to parse the intent prediction signal and generate the optimal service execution instruction based on preset scheduling logic and real-time resource status information. The service scheduling decision core unit includes a static service rule engine that stores a series of "intent-service" mapping rules for handling high-confidence and routine care needs, enabling rapid response scheduling. It also includes a dynamic resource optimization module, which is activated when processing complex or ambiguous intents and is configured to comprehensively evaluate at least one of the following pieces of information to make a scheduling decision: (1) Estimated time and energy costs for each available nursing resource to perform a specific task; (2) Historical preference data or personalized care plans of service recipients; (3) The priority queue of all pending tasks.

[0035] The system's "brain" is a core service scheduling decision-making unit. This unit is crucial for achieving intelligent scheduling. Upon receiving an intent signal, it initiates a complex decision-making process. Internally, this unit contains two parallel decision-making mechanisms to adapt to varying levels of complexity in different scenarios.

[0036] The first mechanism is a static service rule engine. This engine pre-configures a large number of deterministic rules in the form of "IF-THEN" logic. For example, "IF intent is 'drink water' AND confidence > 0.9 AND urgency is 'normal' THEN generate a 'water delivery' service task." This rule engine is mainly used to handle frequent, well-defined, and routine requests with fixed processing procedures. Its advantages are extremely fast response speed, low computational overhead, and the ability to ensure immediate processing of simple requests.

[0037] The second mechanism is the dynamic resource optimization module, which is the more innovative part of this invention. This module is activated when the received intent signal has low confidence, the intent itself is complex, or current nursing resources are strained. This module no longer relies on fixed rules but executes a real-time, multi-objective optimization decision-making process. It comprehensively considers various dynamically changing factors to find the globally optimal scheduling scheme. These factors include at least:

[0038] Resource cost assessment: This module accesses a cost model that stores the estimated costs of using different resources to perform different tasks. Cost can be a multi-dimensional vector, including time costs, energy costs, and even labor costs.

[0039] Service recipient preferences: The system will query the personalized database associated with the service recipient, which may record the elderly person's preference for specific caregivers, acceptance of robot services, or specific care contraindications.

[0040] Global task priority: The module reviews the queue of all pending tasks and sorts them according to their urgency. A new high-urgency task may preempt resources from a low-urgency task that is currently being processed.

[0041] Path and Efficiency: Combining real-time location information, the module calculates the optimal path and estimated time for different resources to reach the service location, and selects the most efficient solution.

[0042] c. A resource status management unit, connected to the service scheduling decision core unit, is configured to monitor and dynamically update a status database containing the availability, location, and workload of various types of nursing resources in real time. The dynamic resource optimization module employs a weighted evaluation algorithm, the mathematical model of which is:

[0043]

[0044] in, For resources Regarding intention The final scheduling score, For cost function, For users The preference function, To execute the efficiency function, These are configurable weighting coefficients.

[0045] The dynamic resource optimization module generates an optimal scheduling solution for the current state, determining "who" and "how" to perform the service. Furthermore, the system includes a resource status management unit. This is a crucial supporting module that provides the decision-making core with a real-time, accurate data foundation. This unit maintains continuous communication with various nursing resources via IoT technology. For nursing staff, their smart terminals periodically report their status and real-time location. For nursing equipment, their built-in sensors automatically report operating status, battery level, location, and other information. All this information is aggregated into a centralized status database and dynamically updated in near real-time. Before making any scheduling decisions, the decision-making core unit must query this database to ensure that its decisions are based on the latest resource conditions.

[0046] d. A task dispatch and feedback unit, connected to the service scheduling decision core unit and the resource status management unit, is configured to issue service execution instructions to designated nursing resources and receive task completion feedback signals after service execution, thereby triggering action on the resource. The task dispatch and feedback unit transmits service execution instructions in the form of structured data to the control system of a mobile smart terminal held by nursing staff or nursing equipment via a wireless communication network.

[0047] This unit is responsible for executing decision results and forming a closed-loop management system. After the service scheduling decision core unit generates a service execution instruction, the instruction is dispatched to this unit. The task dispatch unit accurately distributes the task according to the resource type specified in the instruction, using the corresponding communication protocol. For example, if it's a manual task, a task notification is pushed to the corresponding caregiver's mobile app; if it's an equipment task, a control instruction is sent to the equipment's control system. After the task is completed, the caregiver or equipment sends a "task completed" feedback signal to this unit through the same channel. Upon receiving the feedback, this unit marks the task status as "completed" and immediately notifies the resource status management unit, triggering an update to the status database. For example, it changes the status of the caregiver who just completed the task from "on task" back to "idle," allowing them to be rescheduled. This feedback mechanism ensures closed-loop management of the entire service process and guarantees the real-time nature and accuracy of resource status.

[0048] A proactive nursing service scheduling system and method based on multimodal intent prediction includes the following steps: (1) Receiving steps: Receive a structured intent prediction signal containing the service object identifier and potential intent through the system's communication interface; (2) Decision-making steps: Call the service scheduling decision core unit. First, based on the confidence and clarity of the intent, select to apply the static service rule engine or the dynamic resource optimization module for analysis. In this step, the system queries a real-time updated resource status database to obtain the status of all available nursing manpower and equipment. If the dynamic resource optimization module is activated, the module will execute an optimization algorithm. The goal of the algorithm is to minimize the total scheduling cost of nursing resources while meeting the service quality requirements. (3) Generation steps: Based on the decision analysis results, generate a service execution instruction that includes task type, target resources, and execution parameters; (4) Dispatch step: Send the service execution instruction to the selected nursing resource terminal; (5) Closed-loop feedback step: After receiving the signal that the task is completed, update the status of the corresponding resource in the resource status database to complete a closed-loop scheduling of proactive service.

[0049] The method also includes a logging step to structure and store each intent signal, decision-making process, execution instruction, and task feedback result, providing data support for subsequent service model analysis and system optimization.

[0050] The embodiment will describe in detail the entire process of proactive service scheduling for "delivering water to the elderly" in a typical application scenario to illustrate how the system and method of the present invention work together.

[0051] Scene setting:

[0052] Service recipient: Mr. Wang, a disabled elderly man, lives in Room 301, Building A of the nursing home.

[0053] B) Available nursing resources:

[0054] I. Caregiver Sister Li is currently "available" and is located at the nurses' station on the 3rd floor of Building A.

[0055] II. Caregiver Mr. Zhang is currently "on duty" and is expected to be available in 10 minutes.

[0056] III. Water delivery robot R1, currently "available", located at the charging station on the 3rd floor, with 85% battery.

[0057] C) System time: 15:30 PM.

[0058] Step 1: Receive the intention prediction signal

[0059] The intent recognition system deployed in Mr. Wang's room detected that he stared at the water glass on his bedside table twice within 30 seconds for an extended period. Combining this with sensor data indicating a dry environment, the system generated an intent prediction. The intent signal receiving unit of this invention receives the following structured JSON signal via the internal local area network:

[0060] { "subject_id": "301-Wang", "timestamp": "2025-07-28T15:30:15Z", "intents": [ { "description": "Want to drink water", "confidence": 0.97}, { "description": "Want to take medicine", "confidence": 0.02} ], "urgency": "normal", "location": "Building A - Room 301" }

[0061] Step 2: Service Scheduling Decision

[0062] After receiving the above signal, the system's service scheduling decision-making core unit begins the decision-making process.

[0063] A) Signal Analysis: The unit first analyzes the signal and extracts key information: the service recipient is Grandpa Wang, the highest confidence level of intent is "want to drink water", with a confidence level as high as 0.97, the urgency level is "normal", and the location is room 301.

[0064] B) Decision path selection: Since the intent is clear, the confidence level is high, and it is a routine requirement, the core decision unit prioritizes activating the static service rule engine.

[0065] C) Rule matching: The rule engine performs matching in its rule base and finds a rule: IF intent == "Want to drink water" THEN generate_task "Water delivery service". Therefore, the engine decides to generate a "water delivery service" task.

[0066] Resource Query and Filtering: Before generating specific instructions, the decision-making core unit must query the status database maintained by the resource status management unit to determine the resources available for performing the "water delivery service." The query results are as follows:

[0067] I. Caregiver Sister Li: Status "Idle", Location "Nurse Station, 3rd Floor, Building A", Skills tag includes "Daily Care". Meets the requirements.

[0068] II. Caregiver Mr. Zhang: Status "On Duty". Does not meet the requirements.

[0069] III. Water Delivery Robot R1: Status "Available", Location "3rd Floor Charging Station", Function Label includes "Water Delivery". Meets the requirements.

[0070] D) Activate the dynamic resource optimization module for resource selection: At this point, there are two available resources, Sister Li and Robot R1. The system can be configured to activate the dynamic resource optimization module in this situation to make the optimal choice between the two options.

[0071] E) Cost assessment: The module calculates the cost of the two options.

[0072] Option 1: Ms. Li: The walking distance from the nurses' station to room 301 is 20 meters, estimated to take 1 minute. The task execution is estimated to take 2 minutes. The total time cost is approximately 3 minutes. Labor costs are in standard units.

[0073] II. Option 2: Robot R1: Travel distance from charging station to Room 301 is 25 meters, estimated time 1.5 minutes. Task execution estimated 3 minutes. Total time cost approximately 4.5 minutes. Energy cost is 0.1 kWh.

[0074] F) Preference Query: The module queries Mr. Wang's personalized profile and finds the note "Mr. Wang likes to communicate with people and has a moderate acceptance of robot services." Therefore, the preference function gives Ms. Li a higher weight score.

[0075] G) Final decision: Taking into account time cost and user preference, the evaluation algorithm of the dynamic resource optimization module outputs the final decision: Select caregiver Sister Li to perform this task.

[0076] Step 3: Generating and dispatching service execution instructions

[0077] Based on the decision-making result, the core decision-making unit generates a standardized service execution instruction. This instruction is then sent to the task dispatch and feedback unit.

[0078] The instructions are as follows:

[0079] { "task_id": "TSK-20250728-015", "task_type": "water delivery service", "resource_id": "Caregiver-Li", "subject_id": "301-Wang", "destination": "Building A-301", "instructions": "Please bring a glass of warm water to Mr. Wang in room 301." ", "priority": "regular" }

[0080] The task dispatch unit pushes this instruction to the smartphone app that Sister Li carries with her via Wi-Fi network.

[0081] Step 4: Task Execution and Closed-Loop Feedback

[0082] A) Task Reception and Execution: Ms. Li's mobile app receives a push notification, emits a sound, and displays task details. She clicks "Accept Task." At this time, her app automatically sends a status update signal to the resource status management unit, and Ms. Li's status in the database changes to "Taking Task." Ms. Li then goes to Room 301 to deliver water to Mr. Wang.

[0083] B) Task Completion and Feedback: After Mr. Wang finished drinking the water, Ms. Li clicked "Task Completed" on the APP. The APP then sent a completion signal to the task dispatch and feedback unit again.

[0084] C) System status update: After receiving the completion signal, the feedback unit performs two operations:

[0085] I. Mark the status of task TSK-20250728-015 as "completed" in the task log.

[0086] II. Notify the resource status management unit to update Sister Li's status from "Task" back to "Idle".

[0087] This completes a proactive, intelligent, and closed-loop nursing service scheduling process triggered by intention prediction. The entire process requires no human intervention in decision-making and scheduling, responds quickly, allocates resources rationally, and is fully traceable.

Claims

1. A proactive nursing service scheduling system based on multi-modal intention prediction, characterized by, The system comprises: a. An intention signal receiving unit configured to receive a structured intention prediction signal generated by an external intention recognition system, the signal containing at least a unique identifier of a service object, one or more candidate intentions and their confidence levels; b. A service scheduling decision core unit connected to the intention signal receiving unit, configured to analyze the intention prediction signal and generate an optimal service execution instruction based on pre-set scheduling logic and real-time resource status information; c. A resource status management unit connected to the service scheduling decision core unit, configured to monitor and dynamically update a status database containing the availability, location and workload of multiple types of nursing resources in real time; d. A task assignment and feedback unit connected to the service scheduling decision core unit and the resource status management unit, configured to issue the service execution instruction to the designated nursing resources and receive the task completion feedback signal after the service is executed to trigger the update of the status database in the resource status management unit.

2. The system of claim 1, wherein, The intention prediction signal further contains the urgency level of the intention and the geographic spatial information of the intention occurrence.

3. The system of claim 1, wherein, The service scheduling decision core unit internally includes a static service rule engine that stores a series of "intention-service" mapping rules for processing high-confidence and routine nursing needs to achieve fast response scheduling.

4. The system of claim 1 or 3, wherein, The service scheduling decision core unit also internally includes a dynamic resource optimization module that is activated when processing complex or ambiguous intentions, configured to make scheduling decisions by comprehensively evaluating at least one of the following information: (1) The estimated time cost and energy cost of each available nursing resource to perform a specific task; (2) The historical preference data or personalized care plan of the service object; (3) The priority queue of all pending tasks.

5. The system of claim 4, wherein, The dynamic resource optimization module uses a weighted evaluation algorithm, and the mathematical model of the algorithm is: wherein, is a resource for an intent of final scheduling score, is a cost function, is a user preference function, is an execution efficiency function, is a configurable weight coefficient.

6. The system of claim 1, wherein, The human resource status in the resource status management unit includes at least "idle", "in task" and "in break", and the device resource status includes at least "available", "in operation", "in charging" and "faulty".

7. The system of claim 1, wherein, The task assignment and feedback unit sends the service execution instruction in the form of structured data to the mobile intelligent terminal held by the nursing staff or the control system of the nursing equipment through a wireless communication network.

8. An active nursing service scheduling method based on multi-modal intention prediction, characterized by, The method comprises the following steps: (1) Receiving step: receiving a structured intention prediction signal containing the identifier of the service object and the potential intention through the communication interface of the system; (2) Decision step: calling the service scheduling decision core unit, first selecting to apply the static service rule engine or the dynamic resource optimization module for analysis according to the confidence level and clarity of the intention; in this step, the system queries a real-time updated resource status database to obtain the status of all available nursing manpower and equipment; (3) Generation step: generating a service execution instruction containing the task type, target resource and execution parameters based on the decision analysis result; (4) Assignment step: issuing the service execution instruction to the selected nursing resource terminal; (5) A closed-loop feedback step: after receiving the signal of task completion, the state of the corresponding resource in the resource state database is updated, and the closed-loop scheduling of the active service is completed.

9. The method of claim 8, wherein, In the decision step, if the active resource optimization module is activated, the module will execute an optimization algorithm, the goal of which is to minimize the total scheduling cost of nursing resources under the premise of meeting the quality of service.

10. The method of claim 8, wherein, The method also includes a logging step for structurally storing each intention signal, decision process, execution instruction and task feedback result, providing data support for subsequent service mode analysis and system optimization.

Images