A hospital treatment navigation method based on internet of things technology

Abstract

The present application relates to the technical field of hospital navigation scheduling, in particular to a hospital treatment navigation method based on Internet of Things technology, comprising: a server retrieves a treatment task list containing to-be-executed tasks and physical positions from a hospital information system according to an identity. With the help of an Internet of Things gateway, environmental dynamic data such as the number of queuing diagnosis and treatment nodes, the available state of equipment, the traffic state of key channels and the like collected by hospital sensors are aggregated to construct a dynamic cost atlas containing path travel time and node waiting time. An improved ant colony optimization algorithm is adopted to combine the dynamic cost atlas with the treatment task list to generate a navigation planning scheme with a task sequence and an execution time window, which is issued to a navigation terminal to complete patient treatment guidance. The present method can be realized through a smart phone carried by a patient or a navigation module integrated on various mobile terminals such as electric wheelchairs, electric diagnosis and treatment vehicles and the like.

Description

Technical Field

[0001] This invention relates to the field of hospital navigation and scheduling technology, and in particular to a hospital inpatient navigation method based on Internet of Things (IoT) technology. Background Technology

[0002] The hospital's internal treatment areas have a complex layout, and the treatment process involves multiple sequential steps. Different treatment points are spatially dispersed. Conventional in-hospital navigation services rely solely on static indoor geographic information for route planning, setting routes based only on fixed spatial distances without dynamically adapting to the hospital's real-time operational status. The existing technical architecture lacks the ability to uniformly collect and integrate data from the entire hospital environment, making it difficult to simultaneously obtain information such as patient flow at treatment points, equipment operating conditions, and passageway status. Furthermore, the road network model parameters upon which route planning relies remain fixed over a long period.

[0003] Conventional navigation methods do not incorporate waiting times at treatment nodes and passageway times into the planning and evaluation system. Their path planning dimensions are singular, failing to adapt to fluctuations in access conditions caused by dynamic changes in the hospital environment, and making it difficult to meet the actual business needs of patients requiring multiple tasks consecutively. Traditional path planning calculation models use basic pathfinding logic, focusing only on spatial path optimization without considering the sequential logical relationships of patients' specific treatment tasks, and lacking the ability to constrain arrangements over time.

[0004] To address the technological gaps in multi-source environmental data aggregation and modeling within hospitals, and multi-task time-series navigation planning, it is necessary to leverage an IoT architecture to achieve real-time aggregation of multi-dimensional environmental data within the hospital. Based on real-time environmental parameters, a dynamic road network model integrating travel and waiting times should be constructed. Simultaneously, it is necessary to adapt to specific patient visit sequences, employing optimized planning algorithms to coordinate path and time arrangements, forming a standardized navigation solution adapted to individual patient visit processes, meeting the requirements for orderly progression and rational path planning within the hospital. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a hospital visit navigation method based on Internet of Things (IoT) technology.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a hospital visit navigation method based on Internet of Things (IoT) technology, comprising: The system collects patient identification information by deploying an identification unit on the patient's medical card or mobile terminal and sends it to the navigation management server. Based on the received patient identification information, the navigation management server retrieves the patient's medical task list from the hospital information system. The medical task list includes several tasks to be performed and their associated physical location information. The navigation management server aggregates dynamic environmental data collected in real time from various sensors deployed within the hospital based on an IoT gateway. This dynamic environmental data includes the number of people queuing at each treatment node, the availability of equipment, and the real-time passage status of key channels. The navigation management server constructs a dynamic cost graph containing path travel time and node waiting time based on the dynamic environmental data, and calls an improved navigation planning algorithm to generate a navigation planning scheme containing task sequences and execution time windows based on the dynamic cost graph and the medical visit task list; the improved navigation planning algorithm is an improved ant colony optimization algorithm. The generated navigation plan is sent to the patient's navigation terminal to guide the patient in its execution.

[0007] As a further aspect of the present invention, the navigation management server, based on an IoT gateway, aggregates dynamic environmental data collected in real time from various sensors deployed within the hospital, specifically including: Video analytics sensors are deployed in front of clinics, examination rooms, and pharmacies. These sensors collect and analyze video streams in real time, count the number of people entering specific areas, and output the current number of people in queue. Integrate operational status sensors within medical devices to collect real-time data on device power-on / off status, current operating mode, and estimated idle time, and output device availability status data. Infrared beam sensors or lidar sensors are deployed in building corridors, elevator entrances, and stairwells to collect real-time data on the flow and speed of pedestrians or vehicles passing through specific channel sections, and output real-time traffic status data of key channels. Through the IoT gateway, a unified message queue telemetry transmission protocol is used to asynchronously receive reported data from video analytics sensors, operational status sensors, infrared beam sensors, and lidar sensors. The data reported by various sensors received through the IoT gateway are parsed for format, timestamp aligned, and outlier cleaned to form a structured and time-synchronized dynamic environmental dataset.

[0008] As a further aspect of the present invention, the navigation management server constructs a dynamic cost map containing path travel time and node waiting time based on the environmental dynamic data, specifically including: Using the path intersections and treatment nodes on the hospital's indoor electronic map as vertices and the connecting paths between vertices as edges, an initial topology graph is constructed. Extract real-time queue number data and equipment availability status data for each treatment node from the environmental dynamic dataset; Based on the queuing theory model, real-time queuing data and equipment availability data are converted into the patient's expected waiting time at the corresponding treatment node, and the expected waiting time is used as the dwell cost of the corresponding node vertex. Real-time traffic status data of each key channel is extracted from the environmental dynamic dataset. Combined with the static physical length of the channel, the predicted travel time of pedestrians through the corresponding path segment is calculated, and the predicted travel time is used as the movement cost of the corresponding edge. The dwell cost and movement cost are respectively assigned to the vertices and edges of the initial topology graph to generate a dynamic cost graph that includes dynamic time costs.

[0009] As a further aspect of the present invention, the step of converting real-time queue size data and equipment availability data into the patient's estimated waiting time at the corresponding treatment node based on the queuing theory model specifically involves: Configure the corresponding number of service desks and service intensity parameters for different types of diagnosis and treatment nodes; Use the real-time queue length data acquired by the sensor as the input to the queuing theory model; Convert the available status data of the equipment into the number of valid service desks. If the equipment status is faulty or occupied, subtract the corresponding number from the total number of service desks. By substituting the current queue length, number of effective service counters, and service intensity parameters into the calculation formula in the queuing theory model, the patient's expected waiting time can be calculated.

[0010] As a further aspect of the present invention, the step of invoking the improved navigation planning algorithm to generate a navigation planning scheme containing a task sequence and an execution time window based on the dynamic cost map and the medical visit task list includes: The patient's starting point, each task point in the medical task list, and the ending point are mapped to the corresponding vertices in the dynamic cost graph, which serve as the target nodes that the algorithm needs to access. Initialize the pheromone matrix. The value of each element in the pheromone matrix represents the virtual pheromone concentration of the path connecting two target nodes. The initial concentration is set according to the reciprocal of the sum of the predicted travel time of the corresponding path and the expected waiting time of the target node in the dynamic cost map. Release multiple virtual ants. Each virtual ant starts from the starting point and selects the next target node to visit according to the state transition rules. The state transition rules comprehensively consider the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node. Once a virtual ant has completed the path construction from the starting point, visited all task points, and finally reached the destination, the total dynamic cost of the path is calculated. The total dynamic cost is the sum of the movement costs of all nodes on the path and the dwell costs of all visited nodes. Based on the total dynamic cost of the virtual ant's path construction, the corresponding values ​​of the pheromone in the pheromone matrix are updated according to the rule that the lower the cost, the more pheromone is released.

[0011] After multiple virtual ants complete the iteration, the path with the lowest total dynamic cost in the historical iterations is selected as the optimal solution output. The resulting path is the navigation planning scheme containing the task sequence. Based on the task sequence in the navigation planning scheme, and combined with the predicted travel time between each task point and the estimated waiting time at each node, the estimated time for the patient to arrive at each task point is calculated, forming an execution time window.

[0012] As a further aspect of the present invention, the state transition rule comprehensively considers the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node, specifically as follows: Calculate the transition probability from the current node to each unvisited candidate node. The transition probability is positively correlated with the pheromone concentration on the path connecting the current node and the candidate node, positively correlated with the inverse of the real-time movement cost of the corresponding path in the dynamic cost graph, and positively correlated with the inverse of the real-time residence cost of the candidate node in the dynamic cost graph. Based on the calculated transition probabilities of each candidate node, the roulette wheel method is used to select the next node to be actually visited.

[0013] As a further aspect of the present invention, after multiple virtual ants complete the iterations and select the path with the lowest total dynamic cost in the historical iterations as the optimal solution output, the method further includes performing a time feasibility verification on the navigation planning scheme, specifically: Extract the task sequence and the predicted travel time between each task point from the navigation planning scheme, as well as the estimated waiting time at each task point; Obtain the appointment time limits or latest completion time requirements associated with each medical task from the hospital information system; Based on the task sequence, starting from the current system time, accumulate the travel time and waiting time to calculate the estimated arrival time and estimated departure time of each task point; The estimated departure time for each task point is compared with the latest completion time requirement for the corresponding task obtained from the hospital information system; If the estimated departure time of any task point is later than its latest completion time requirement, the current navigation planning scheme is deemed infeasible, and the task sequence needs to be readjusted or an alarm needs to be triggered. If the estimated departure time of all task points is no later than their latest completion time requirement, the navigation planning scheme is deemed feasible and the output is confirmed.

[0014] As a further aspect of the present invention, the step of sending the generated navigation plan to the navigation terminal held by the patient and guiding the patient to execute it specifically includes: The navigation management server encodes the navigation planning scheme, which includes task sequences and execution time windows, into a specific data protocol format; The encoded navigation plan is sent to the patient's navigation terminal via wireless local area network; The navigation terminal decodes the received navigation plan and displays the current task, the next task target point, and the arrow guide path from the current location to the next target point on the graphical interface; The navigation terminal obtains its real-time location through the built-in indoor positioning module and compares it with the planned path. If it detects that the patient deviates from the planned path beyond the allowable error range, it will provide correction prompts through voice or vibration. Once the patient reaches a task point and completes the corresponding task, the navigation terminal receives confirmation of task completion through interface interaction or automatic sensing, and sends this status back to the navigation management server.

[0015] As a further aspect of the present invention, the method further includes: During the guided execution process, the system continuously monitors specific environmental dynamic data change events reported by the IoT gateway. When the monitored events meet the preset replanning trigger conditions, a new navigation planning scheme is regenerated and switched based on the current real-time location of the patient. During the boot process, continuous monitoring of specific environmental dynamic data change events reported by the IoT gateway includes: Set event monitoring rules in the navigation management server. The event monitoring rules include: events such as the number of people queuing at a specific treatment node exceeding a preset threshold, events such as the passage speed of a key channel being lower than a preset threshold, and events such as medical equipment changing from an available state to a faulty state. The navigation management server subscribes to specific topic messages published by the IoT gateway and continuously receives dynamic environmental data streams. The received environmental dynamic data stream is parsed in real time and matched with the event monitoring rules; When the data parsed in real time meets the conditions of any event monitoring rule, it is determined that an environmental dynamic data change event has been detected, and the type, location and time of the corresponding event are recorded.

[0016] As a further aspect of the present invention, when the detected event meets the preset replanning trigger condition, a new navigation planning scheme is regenerated and switched based on the current real-time location of the patient, specifically including: A replanning trigger condition is predefined, wherein the replanning trigger condition is: the detected event causes the increase in the path movement cost to the corresponding target node in the current navigation planning scheme to exceed the tolerance limit; After detecting environmental dynamic data change events, assess the impact of the event on the currently executing navigation planning scheme, and calculate the expected waiting time increment of affected nodes or the movement cost increment of affected paths; The calculated time increment is compared with the preset tolerance limit. If the time increment exceeds the tolerance limit, the replanning trigger condition is determined to be met. Obtain the real-time location information reported by the patient's navigation terminal, and use the real-time location information as the starting point for new path planning; Based on the latest environmental dynamic data, the dynamic cost map is reconstructed, and an improved navigation planning algorithm is invoked to generate a new navigation planning scheme with a new starting point and a list of remaining unfinished tasks. The new navigation plan is sent to the patient's navigation terminal, which then interrupts the current guidance and switches to execute the new navigation plan.

[0017] Compared with the prior art, the advantages and positive effects of the present invention are as follows: By leveraging IoT gateways to achieve unified aggregation of data from various sensors within the hospital, this system integrates multi-dimensional information such as queue lengths at treatment nodes, equipment availability, and real-time traffic status of key passageways. Based on real-time environmental data, a dynamic cost map is constructed that fuses path travel time and node waiting time. This breaks free from the static constraints of a fixed geographical road network, allowing road network cost parameters to dynamically change with the hospital's operational status. It expands the evaluation dimensions in the path planning process, incorporating both travel and waiting costs as constraints, and refining the quantitative expression of the hospital's road network status.

[0018] An improved ant colony optimization algorithm combined with a dynamic cost graph and a patient visit task list is used for computation and solution. During the planning process, the inherent logic of patient visit operations is followed to arrange the order of tasks and match corresponding execution time windows for each treatment step. This breaks through the limitation of using physical distance as the sole planning basis, incorporating spatial path, task order, and time constraints into the same computational framework to form an overall planning result under multi-dimensional constraints, enriching the structured components of the navigation scheme.

[0019] The complete planning scheme is directly pushed to the navigation terminal to guide the process, using standardized task sequences and time windows to regulate the pace of medical visits and standardize the movement of personnel within the hospital and the arrival time of treatment nodes. This reduces the disorder of the process caused by single-path guidance, unifies and adapts the complete flow logic of individual medical tasks, and ensures that the distribution of personnel flow within the hospital matches the operational capacity of medical resources, maintaining the stability of the overall operation rhythm of medical services within the hospital. Attached Figure Description

[0020] Figure 1 This is a state diagram of a hospital visit navigation method based on Internet of Things technology as described in this invention. Figure 2 A flowchart for the collection and aggregation of dynamic environmental data; Figure 3 A flowchart for constructing a dynamic cost graph. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0022] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] See Figure 1A hospital visit navigation method based on Internet of Things (IoT) technology is proposed, with the following overall implementation scheme: The system collects patient identification information via an identification unit deployed on the patient's medical card or mobile terminal and sends this information to a navigation management server. Based on the received patient identification information, the navigation management server retrieves the patient's visit task list from the hospital information system. This task list contains several tasks to be executed and their associated physical location information. The navigation management server aggregates real-time environmental dynamic data collected by various sensors deployed within the hospital via an IoT gateway. This environmental dynamic data includes the number of people queuing at each treatment node, equipment availability status, and real-time passage status of key channels. Based on the aggregated environmental dynamic data, the navigation management server constructs a dynamic cost graph containing path travel time and node waiting time, and calls an improved navigation planning algorithm (an improved ant colony optimization algorithm). Based on the dynamic cost graph and the visit task list, it generates a navigation planning scheme containing task sequences and execution time windows. The generated navigation planning scheme is then distributed to the patient's navigation terminal to guide the patient's execution. The generated navigation plan is sent to the patient's navigation terminal to guide the patient in execution. The navigation terminal can be the patient's own smartphone, tablet, or other mobile terminal, or a dedicated navigation device provided by the hospital. This method can also be applied to mobile vehicles such as electric wheelchairs and electric medical vehicles integrated with navigation modules, providing navigation guidance to patients or medical staff through onboard displays, voice prompts, etc., thus serving people with mobility impairments or special medical scenarios. The core function of the navigation terminal is to receive and execute the navigation plan; its specific form should not constitute a limitation of this invention.

[0024] In one embodiment of the present invention, see [reference] Figure 2 Video analytics sensors are deployed in front of clinics, examination rooms, and pharmacies to collect and analyze video streams in real time, count the number of people entering specific areas, and output current queue length data. Operating status sensors are integrated into medical equipment to collect real-time data on equipment power-on / off status, current operating mode, and estimated idle time, outputting equipment availability data. Infrared beam sensors or lidar sensors are deployed in building corridors, elevator entrances, and stairwells to collect real-time data on pedestrian or vehicle flow and speed across specific passageways, outputting real-time traffic status data for key passageways. The navigation management server, via an IoT gateway, uses a unified message queue telemetry transmission protocol to asynchronously receive data reported from video analytics sensors, operating status sensors, infrared beam sensors, and lidar sensors. The navigation management server performs format parsing, timestamp alignment, and outlier cleanup on the various sensor reports received through the IoT gateway, forming a structured and time-synchronized dynamic environmental dataset.

[0025] In practical implementation, video analytics sensors are deployed in front of clinics, examination rooms, and pharmacies. These sensors collect and analyze video streams in real time, counting the number of people entering specific areas and outputting the current queue size. For example, in front of the electrocardiogram (ECG) room, the sensor analyzes human silhouettes and movement trajectories in the video stream to count the number of people entering the ECG waiting area and outputs a queue size of 15. In front of the blood drawing room, the sensor outputs a queue size of 8. This data comparison shows the difference in congestion levels at different treatment points. In some embodiments, the video analytics sensors use a combination of background subtraction and human detection algorithms to extract the number of people entering the area from the video stream. The sensors encapsulate the count data in JSON format and report it through an IoT gateway. The data reporting frequency is once per second, ensuring the real-time nature of the queue size data. The operational status sensor is integrated inside the medical device. The operational status sensor collects the device's power-on / off status, current operating mode, and estimated idle time in real time, and outputs the device's availability status data. For example, in a CT scanner, the operational status sensor monitors the power status, current scanning mode, and next scheduled time. The operational status sensor outputs the device's availability status as "busy" with an estimated idle time of 30 minutes. In an ultrasound device, the operational status sensor outputs the device's availability status as "idle" with an estimated idle time of 0 minutes. The data comparison reflects the device's busy / idle status. Infrared beam sensors or lidar sensors are deployed in building corridors, elevator entrances, and stairwells. These sensors collect real-time data on the flow and speed of pedestrians or vehicles passing through specific passage sections, outputting real-time traffic status data for key passages. For example, in the main corridor of the outpatient hall, lidar sensors scan pedestrians passing through the corridor and calculate the pedestrian flow as 120 people per minute with an average speed of 0.8 meters per second. In the emergency passage, infrared beam sensors output a pedestrian flow of 200 people per minute with an average speed of 1.2 meters per second. The data comparison reveals the differences in traffic efficiency between different passages.

[0026] The IoT gateway employs a unified message queue telemetry protocol to asynchronously receive reported data from video analytics sensors, operational status sensors, infrared beam sensors, and LiDAR sensors. In practice, the IoT gateway subscribes to sensor data topics. Video analytics sensors publish data to the "sensor / video / queue" topic, operational status sensors publish data to the "sensor / device / status" topic, and infrared beam sensors or LiDAR sensors publish data to the "sensor / passage / traffic" topic. The IoT gateway receives and caches messages from these topics. The message queue telemetry protocol ensures lightweight and asynchronous data transmission, supporting high-concurrency sensor connections. Essentially, the IoT gateway encapsulates different sensor data using a unified message format; for example, the message payload includes sensor identifier, timestamp, and data value fields. The IoT gateway performs preliminary verification of received messages, discarding those with incorrect formats. The navigation management server performs format parsing, timestamp alignment, and outlier removal on various sensor-reported data received through the IoT gateway, forming a structured and time-synchronized dynamic environmental dataset. Format parsing converts JSON-formatted data into an internal data structure, timestamp alignment synchronizes the timestamps of different sensor data to a unified time base, and outlier removal removes unreasonable data points. In some embodiments, timestamp alignment uses a network time protocol to synchronize all sensor clocks, calibrating the timestamps of sensor data to the navigation management server's system time. The accuracy of timestamp alignment is controlled at the millisecond level to ensure the consistency of the data time series. Optionally, format parsing supports multiple data format conversions, including mapping from JSON to an internal data model. The internal data model includes fields such as queue size, device status, and passage status, with each field associated with a parsed value and a timestamp. Outlier removal uses statistical methods to identify and filter abnormal data, for example, defining a formula to detect outliers. Where: x represents the data point reported by the sensor, μ represents the mean of recent data points, and σ represents the standard deviation of recent data points. When the absolute value of z exceeds a preset threshold, data point x is considered an outlier and is removed. For example, for queuing data, the preset threshold is set to 3, corresponding to the 3σ principle of normal distribution. If the number of people queuing reported by the video analysis sensor is 100, and the recent mean μ is 20 and the standard deviation σ is 5, then the z value is 16, exceeding the threshold, and this data point is removed. Optionally, outlier cleaning also uses a sliding window method to calculate μ and σ. The sliding window contains the most recent 100 data points to ensure dynamic updates of the mean and standard deviation to adapt to changes in data distribution. The navigation management server organizes the cleaned data in chronological order to form an environmental dynamic dataset. The environmental dynamic dataset contains fields such as queuing number, equipment status, and passage status. Each field is associated with a unified timestamp. The environmental dynamic dataset is stored in the database in time series form and supports querying by time range. In practice, the environmental dynamic dataset is stored in a time-series database for the navigation management server to use in building a dynamic cost map. The time-series database optimizes the writing and reading efficiency of time-series data, and the update frequency of the environmental dynamic dataset is consistent with the sensor reporting frequency to maintain the real-time nature of the data.

[0027] In one embodiment of the present invention, see [reference] Figure 3 An initial topology graph is constructed using path intersections and treatment nodes on the hospital's indoor electronic map as vertices and connecting paths between vertices as edges. Real-time queuing data and equipment availability data for each treatment node are extracted from the environmental dynamic dataset. Based on queuing theory, the real-time queuing data and equipment availability data are converted into the estimated waiting time for patients at the corresponding treatment node, and this estimated waiting time is used as the dwell cost of the corresponding node vertex. Real-time traffic status data for each key passage is extracted from the environmental dynamic dataset, and combined with the static physical length of the passage, the predicted travel time for pedestrians through the corresponding path segment is calculated, and this predicted travel time is used as the movement cost of the corresponding edge. The dwell cost and movement cost are weighted onto the vertices and edges of the initial topology graph, respectively, generating a dynamic cost graph containing dynamic time costs. For different types of medical treatment nodes, the corresponding number of service stations and service intensity parameters are configured. The real-time queuing number data obtained by the sensor is used as the input of the current queue length in the queuing theory model. The available status data of the equipment is converted into the number of effective service stations. If the equipment status is faulty or occupied, the corresponding number is subtracted from the total number of service stations. The current queue length, the number of effective service stations and the service intensity parameters are substituted into the calculation formula in the queuing theory model to solve for the patient's expected queuing waiting time.

[0028] In practical implementation, the navigation management server constructs an initial topology graph using path intersections and treatment nodes on the hospital's indoor electronic map as vertices and connecting paths between vertices as edges. For example, path intersections include corridor turning points, elevator shafts, and stairwells; treatment nodes include clinic 201, the radiology department, and the blood drawing window in the laboratory. The initial topology graph has a total of 120 vertices and 200 edges, each edge associated with a static physical length attribute, such as the corridor connecting the outpatient hall and clinic area A having a length of 50 meters. The navigation management server extracts real-time queuing data and equipment availability data for each treatment node from the environmental dynamic dataset. For example, the real-time queuing data for clinic 201 is 5 people, the equipment availability data for the ultrasound examination room is "idle," and the real-time queuing data for the blood drawing window in the laboratory is 12 people. Data comparison shows that the load status of different treatment nodes is different. In some embodiments, the navigation management server obtains real-time queue number data and device availability status data by querying the latest timestamp record of the environmental dynamic dataset. The environmental dynamic dataset is stored in key-value format, where the key is the treatment node identifier and the value is the corresponding data. The navigation management server retrieves the corresponding data based on the treatment node identifier.

[0029] Based on a queuing theory model, real-time queue size data and equipment availability data are converted into the estimated waiting time for patients at the corresponding treatment node, and this estimated waiting time is used as the dwell cost of the corresponding node vertex. For different treatment node types, corresponding service station numbers and service intensity parameters are configured. Real-time queue size data acquired by sensors is used as the current queue length input in the queuing theory model, and equipment availability data is converted into the number of effective service stations. For example, for a clinic-type node, the configured number of service stations is 1, the service intensity parameter is an average service rate of 4 people per hour, the current queue length is 5 people, the equipment status is available, and the number of effective service stations is 1. In specific implementation, the M / M / c queuing model is used to calculate the estimated queuing waiting time. The calculation formula in the queuing theory model is: in: This indicates the patient's estimated waiting time in the queue. Indicates the current queue length. Indicates the number of valid service counters. This represents the average service rate in the service intensity parameters. It can be understood that if a device is faulty or occupied, the corresponding number is subtracted from the total number of service counters. For example, if an inspection room has 2 devices, and the operating status sensor shows that 1 device is faulty, then the effective number of service counters is [number missing]. The value is 1. Substituting the current queue length, number of valid service counters, and service intensity parameters into the calculation formula in the queuing theory model, the patient's estimated waiting time is calculated. For example, for clinic 201... The estimated queuing time is calculated based on the number of people per hour. This value, representing the dwell time at vertex 201 in clinic, is 1.25 hours. Optional, the average service rate in the service intensity parameter. Based on historical service records, different node types were identified. Different values, such as those at the blood drawing window. The appointment registration desk can accommodate 20 people per hour. The limit is 30 people per hour.

[0030] The navigation management server extracts real-time traffic status data for each key passage from the environmental dynamic dataset. Combined with the static physical length of the passage, it calculates the predicted travel time for pedestrians traversing the corresponding path segment and uses this predicted travel time as the movement cost of the corresponding edge. For example, for a 50-meter-long corridor connecting the outpatient hall and clinic area A, the real-time traffic status data extracted from the environmental dynamic dataset shows an average pedestrian speed of 0.5 meters per second. Therefore, the predicted travel time is 100 seconds, and this value is assigned as the movement cost of that edge. In some embodiments, the predicted travel time T is calculated by dividing the static physical length L of the passage by the average speed V extracted from the real-time traffic status data, i.e. For example, if a stairwell passage is 20 meters long (L) and the average speed (V) is 0.3 meters per second, the predicted travel time (T) is approximately 66.7 seconds. Data comparison shows that the movement cost on different sides varies significantly depending on the traffic conditions. For instance, the movement cost on a main road is 100 seconds, while the movement cost on an emergency exit is 200 seconds. It can be understood that the average speed (V) in the real-time traffic status data is calculated from the flow and speed data reported by infrared beam sensors or lidar sensors. The calculation uses the average speed of all pedestrians passing through in the most recent minute to ensure that the movement cost reflects the immediate traffic conditions.

[0031] In implementation, dwell cost and movement cost are weighted onto the vertices and edges of the initial topology graph, respectively, generating a dynamic cost graph containing dynamic time costs. The navigation management server assigns a dwell cost to each vertex attribute and a movement cost to each edge attribute of the initial topology graph, with both dwell and movement costs measured in time units (seconds). For example, the dwell cost of vertex "Clinic 201" is 4500 seconds (1.25 hours), and the movement cost of edge "Corridor AB" is 100 seconds. The dynamic cost graph is stored and maintained in memory using an adjacency list data structure, supporting fast lookup of the cost of a path between any two points. Optionally, the cost values ​​in the dynamic cost graph are periodically refreshed based on updates to the dynamic environmental dataset, with a refresh frequency of once per minute, ensuring that the cost values ​​reflect the dynamic changes in the hospital environment. When generating a navigation planning scheme, the navigation management server queries the dynamic cost graph to obtain real-time vertex dwell costs and edge movement costs, which are used to improve the computational input of the navigation planning algorithm.

[0032] In one embodiment of the invention, the patient's starting point, each task point in the medical task list, and the destination are mapped to corresponding vertices in a dynamic cost graph, serving as target nodes to be visited by the algorithm. A pheromone matrix is ​​initialized, where each element represents the virtual pheromone concentration of the path connecting two target nodes. The initial concentration is set based on the reciprocal of the sum of the predicted travel time of the corresponding path in the dynamic cost graph and the expected waiting time of the target node. Multiple virtual ants are released, each starting from the starting point and selecting the next target node to visit according to a state transition rule. This rule comprehensively considers the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node. After a virtual ant completes the path construction from the starting point, visits all task points, and finally reaches the destination, the total dynamic cost of the path is calculated. The total dynamic cost is the sum of all movement costs on the path and the dwell costs of all visited nodes. Based on the total dynamic cost of the virtual ant's path construction, the corresponding value of the ant's path in the pheromone matrix is ​​updated according to the rule that the lower the cost, the more pheromone is released. After multiple virtual ants complete the iterations, the path with the lowest total dynamic cost in the historical iterations is selected as the optimal solution output. The resulting path is the navigation planning scheme containing the task sequence. Based on the task sequence in the navigation planning scheme, combined with the predicted travel time between each task point and the estimated waiting time of each node, the estimated time for the patient to arrive at each task point is calculated, forming an execution time window. In the state transition rule, the transition probability from the current node to each unvisited candidate node is calculated. This transition probability is positively correlated with the pheromone concentration on the path connecting the current node and the candidate node, positively correlated with the reciprocal of the real-time movement cost of the corresponding path in the dynamic cost graph, and positively correlated with the reciprocal of the real-time residence cost of the candidate node in the dynamic cost graph. Then, based on the calculated transition probabilities of each candidate node, a roulette wheel method is used to select the next node to be actually visited.

[0033] In implementation, the navigation management server maps the patient's starting point, each task point in the patient's task list, and the destination to corresponding vertices in the dynamic cost graph, serving as target nodes that the algorithm needs to access. For example, the patient's starting point is mapped to the "Outpatient Hall Entrance" vertex; the patient's task list includes three task points: "Clinic Room 201," "Laboratory Blood Drawing Window," and "Pharmacy Window 3," each mapped to a vertex with the same name; and the destination is mapped to the "Exit" vertex, for a total of five target nodes that need to be accessed. The navigation management server initializes the pheromone matrix, where each element represents the virtual pheromone concentration of the path connecting two target nodes. The initial concentration of the pheromone matrix is ​​set based on the reciprocal of the sum of the predicted travel time of the corresponding path in the dynamic cost graph and the expected waiting time of the target node. For example, the predicted travel time for the path connecting "Outpatient Hall Entrance" and "Clinic Room 201" in the dynamic cost graph is 180 seconds, and the predicted waiting time for the "Clinic Room 201" vertex is 300 seconds. The sum of the two is 480 seconds, and its reciprocal is 1 / 480. This value is used as the initial concentration value of the corresponding element in the pheromone matrix. The dimension of the pheromone matrix is ​​the number of target nodes multiplied by the number of target nodes. It is a symmetric matrix with diagonal elements having a value of 0.

[0034] In implementation, the navigation management server releases multiple virtual ants. Each virtual ant starts from the origin and selects the next target node to visit according to state transition rules. These rules comprehensively consider the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node. Specifically, the transition probability from the current node to each unvisited candidate node is calculated. This probability is positively correlated with the pheromone concentration on the path connecting the current node and the candidate node, positively correlated with the reciprocal of the real-time movement cost of the corresponding path in the dynamic cost graph, and positively correlated with the reciprocal of the real-time dwell cost of the candidate node in the dynamic cost graph. For example, if the current node is "Outpatient Hall Entrance," and unvisited candidate nodes include "Clinic 201," "Laboratory Blood Drawing Window," "Pharmacy Window 3," and "Exit," the navigation management server calculates a transition probability value for each candidate node. The formula for calculating the transition probability is: in: This represents the transition probability of moving from the current node i to the candidate node j. This represents the pheromone concentration along the path connecting node i and node j. It is heuristic information, and its value is the reciprocal of the real-time movement cost from node i to node j in the dynamic cost graph. It is a node heuristic, and its value is the reciprocal of the real-time residence cost of candidate node j in the dynamic cost graph. It is a pheromone factor. It is a heuristic factor for mobile cost. It is a heuristic factor for the cost of residence. This represents the set of candidate nodes that have not yet been visited. This is understandable. These are preset parameters used to adjust the relative importance of pheromone concentration, movement cost, and dwell cost during state transitions. For example, setting... This indicates that the impact weight of movement cost is higher than that of pheromone and dwell cost. Based on the calculated transition probabilities of each candidate node, a roulette wheel method is used to select the next node to be actually visited. The roulette wheel method assigns a sector region proportional to the transition probability to each candidate node, and then generates a random number between 0 and 1. The node corresponding to the sector region where the random number falls is selected as the next visited node. For example, see Table 1 for the calculation of candidate nodes.

[0035] Table 1: Calculation Table of State Transition Probability In some embodiments, the number of virtual ants is set to 50. Each virtual ant independently constructs a path according to the state transition rules. A virtual ant starts from the starting point, visits all task points, and finally reaches the destination, thus forming a complete access path. After a virtual ant completes the path construction from the starting point to the destination, the total dynamic cost of the path is calculated. The total dynamic cost is the sum of the movement costs of all nodes on the path and the dwell costs of all visited nodes. For example, a virtual ant constructs a path as "Outpatient Hall Entrance" -> "Clinic 201" -> "Laboratory Blood Drawing Window" -> "Pharmacy Window No. 3" -> "Exit". According to the dynamic cost graph, the movement cost from "Outpatient Hall Entrance" to "Clinic 201" is 180 seconds, from "Clinic 201" to "Laboratory Blood Drawing Window" is 150 seconds, from "Laboratory Blood Drawing Window" to "Pharmacy Window No. 3" is 100 seconds, and from "Pharmacy Window No. 3" to "Exit" is 80 seconds. The dwell cost at node "Clinic 201" is 300 seconds, at "Laboratory Blood Drawing Window" is 600 seconds, and at "Pharmacy Window No. 3" is 120 seconds. The total dynamic cost is (180+150+100+80)+(300+600+120)=1530 seconds. It is understandable that the dwell costs at the starting and ending points of the path are not included in the total dynamic cost.

[0036] Based on the total dynamic cost of the virtual ant's path construction, and following the rule that lower costs release more pheromones, the corresponding values ​​of the ant's path in the pheromone matrix are updated; the pheromone update formula is: in: It is the original pheromone concentration along path (i,j). It is the pheromone volatile factor (0 < ρ < 1). This is the sum of pheromones released by all ants along path (i,j) in this iteration. For a single ant k, the amount of pheromones it releases along path (i,j) is... Total dynamic cost of constructing the path with it Inversely proportional, that is Where Q is the pheromone intensity constant. For example, the total dynamic cost of an ant constructing a path. Given a time interval of 1530 seconds and Q=1000, the increment of pheromones released by the ant on each edge it traverses is... The value is approximately 1000 / 1530 ≈ 0.6536. The sum of the pheromone increments for all 50 ants on each edge in this iteration is... The pheromone evaporation factor ρ is set to 0.5, meaning that half of the original pheromone will evaporate after each iteration, and this, combined with the pheromone released by the ants in the current iteration, forms a new pheromone concentration. After multiple virtual ants complete the iterations, the path with the lowest total dynamic cost in the historical iterations is selected as the optimal solution output. The resulting path is the navigation planning scheme containing the task sequence. For example, if the algorithm runs 10 iterations, releasing 50 ants in each iteration, and records the path with the lowest total dynamic cost from all 500 constructed paths, assuming this path is "Outpatient Hall Entrance" -> "Pharmacy Window 3" -> "Laboratory Blood Drawing Window" -> "Clinic 201" -> "Exit", its total dynamic cost is 1420 seconds. This path is output as the task sequence of the navigation planning scheme: "Pharmacy Window 3, Laboratory Blood Drawing Window, Clinic 201".

[0037] Optionally, based on the task sequence in the navigation planning scheme, and combining the predicted travel time between each task point with the estimated waiting time at each node, the estimated time for the patient to arrive at each task point is calculated, forming an execution time window. The navigation management server obtains the predicted travel time between adjacent nodes in the task sequence and the estimated waiting time for each task node from the dynamic cost map, and accumulates them starting from the current system time. For example, if the current system time is 09:00:00 and the task sequence is "Pharmacy Window 3, Laboratory Blood Drawing Window, Clinic 201", the predicted travel time from the starting point to "Pharmacy Window 3" is 200 seconds, so the estimated arrival time at "Pharmacy Window 3" is 09:03:20. The estimated waiting time at "Pharmacy Window 3" is 120 seconds, so the estimated departure time is 09:05:20. The predicted travel time from "Pharmacy Window 3" to "Laboratory Blood Drawing Window" is 100 seconds, so the estimated arrival time at "Laboratory Blood Drawing Window" is 09:07:00, and so on, forming the estimated arrival and departure time windows for each task point, i.e., the execution time window.

[0038] In one embodiment of the present invention, the task sequence and predicted travel time between each task point, as well as the estimated waiting time at each task point, are extracted from the navigation planning scheme. The appointment time limit or latest completion time requirement associated with each treatment task is obtained from the hospital information system. Following the task sequence order, starting from the current system time, the travel time and waiting time are accumulated to calculate the estimated arrival time and estimated departure time for each task point. The estimated departure time of each task point is compared with the latest completion time requirement corresponding to the task obtained from the hospital information system. If the estimated departure time of any task point is later than its latest completion time requirement, the current navigation planning scheme is deemed infeasible, and the task sequence needs to be readjusted or an alarm triggered. If the estimated departure time of all task points is not later than their latest completion time requirement, the navigation planning scheme is deemed feasible, and the output is confirmed. The navigation management server encodes the navigation planning scheme, which includes the task sequence and execution time window, into a specific data protocol format and distributes the encoded navigation planning scheme to the navigation terminal held by the patient via a wireless local area network. The navigation terminal decodes the received navigation plan and displays the current task, the next task target point, and arrow guidance paths from the current location to the next target point on a graphical interface. The navigation terminal obtains its real-time location through a built-in indoor positioning module and compares it with the planned path. If it detects that the patient has deviated from the planned path beyond the allowable error range, it provides correction prompts via voice or vibration. When the patient reaches a task point and completes the corresponding task, the navigation terminal receives confirmation of task completion through interface interaction or automatic sensing and feeds this status back to the navigation management server.

[0039] In practical implementation, the navigation management server extracts the task sequence and predicted travel time between each task point, as well as the estimated waiting time at each task point, from the navigation planning scheme. For example, if the task sequence in the navigation planning scheme is "Pharmacy Window 3, Laboratory Blood Drawing Window, Clinic 201", the predicted travel time from "Outpatient Hall Entrance" to "Pharmacy Window 3" is 200 seconds, from "Pharmacy Window 3" to "Laboratory Blood Drawing Window" is 100 seconds, and from "Laboratory Blood Drawing Window" to "Clinic 201" is 150 seconds. The estimated waiting time at the task point "Pharmacy Window 3" is 120 seconds, the estimated waiting time at "Laboratory Blood Drawing Window" is 600 seconds, and the estimated waiting time at "Clinic 201" is 300 seconds. The navigation management server obtains the appointment time limits or latest completion time requirements associated with each treatment task from the hospital information system, and the hospital information system returns structured data through a data interface. For example, in the patient's medical task list, the latest completion time requirement for the test items associated with the "Blood Drawing Window of the Laboratory Department" is "10:30 on the same day", and the time limit for doctor's consultation appointment associated with "Clinic Room 201" is "10:00 to 11:00 on the same day". Data comparison shows that different tasks have different time constraints.

[0040] Following the task sequence, starting from the current system time, the estimated arrival and departure times for each task point are calculated by accumulating the movement and waiting times. The current system time is set to 09:00:00. The calculation process is as follows: Moving from the starting point to "Pharmacy Window 3" takes 200 seconds, arriving at 09:03:20; waiting at "Pharmacy Window 3" for 120 seconds results in departure at 09:05:20; Moving from "Pharmacy Window 3" to "Laboratory Blood Drawing Window" takes 100 seconds, arriving at 09:07:00; waiting at "Laboratory Blood Drawing Window" for 600 seconds results in departure at 09:17:00; Moving from "Laboratory Blood Drawing Window" to "Clinic Room 201" takes 150 seconds, arriving at 09:19:30; waiting at "Clinic Room 201" for 300 seconds results in departure at 09:24:30. In some embodiments, the time calculation formula is expressed as: in: This represents the estimated arrival time to the k-th task point. This represents the estimated waiting time for the k-th task point. This represents the estimated departure time from the k-th task point. This represents the predicted travel time from the k-th task point to the (k+1)-th task point, where the starting point of the sequence is... Add the travel time from the starting point to the first task point to the current system time. This formula can be understood as iteratively calculating the arrival and departure times of each task point to form a complete execution time window.

[0041] The estimated departure time for each task point is compared with the latest completion time requirement for the corresponding task obtained from the hospital information system. For example, the estimated departure time for the task point "Blood Drawing Window in the Laboratory" is 09:17:00, and its latest completion time requirement is "10:30 on the same day." The estimated departure time is earlier than the latest completion time, thus meeting the requirement. The estimated departure time for the task point "Clinic 201" is 09:24:30, and its appointment time limit is "10:00 to 11:00 on the same day." The estimated departure time falls within this time window, thus meeting the requirement. In practice, the time comparison operation converts the time into seconds starting from midnight of the current day for comparison, avoiding differences in date and time format parsing. For example, "10:30" is converted to 37800 seconds, and "09:24:30" is converted to 33870 seconds. Comparing 33870 < 37800, it is determined that the requirement is met. The navigation management server executes judgment logic based on the comparison results. If the estimated departure time of any task point is later than its latest completion time requirement, the current navigation planning scheme is deemed infeasible, and the task sequence needs to be readjusted or an alarm triggered. For example, if the estimated departure time of "Clinic 201" is calculated to be 11:30:00, while the latest completion time requirement is 11:00:00, the scheme is deemed infeasible. If the estimated departure time of all task points is not later than their latest completion time requirement, the navigation planning scheme is deemed feasible, and the output is confirmed. Refer to Table 2, which shows a feasible example of calculation and comparison results: Table 2: Time Feasibility Verification Table for Navigation Planning Schemes The navigation management server encodes the navigation planning scheme, which includes task sequences and execution time windows, into a specific data protocol format. For example, the navigation management server uses JSON-LD format to encode the scheme. The data packet contains fields such as "patientID", "taskSequence", and "timeWindows". The "taskSequence" field value is an array ["Pharmacy Window 3", "Laboratory Blood Drawing Window", "Clinic Room 201"], and the "timeWindows" field value is an array containing the estimated arrival and departure timestamps for each task point. In some embodiments, the encoding protocol uses ProtocolBuffers format to improve transmission efficiency. The .proto file defines the message structure of the navigation scheme, and the navigation management server serializes the internal data structure into a binary stream. The encoded navigation planning scheme is distributed to the navigation terminal held by the patient via a wireless LAN. Wireless LAN access points are deployed in various areas of the hospital, and the navigation management server pushes the data packet to the network address bound to the navigation terminal via the TCP / IP protocol.

[0042] The navigation terminal decodes the received navigation plan and displays the current task, the next task target point, and an arrow indicating the path from the current location to the next target point on a graphical interface. For example, after decoding, the top of the graphical interface displays "Current Task: Go to Pharmacy Window 3," the middle displays a hospital floor plan, and the location of "Pharmacy Window 3" is marked with a flashing dot on the floor plan. A highlighted arrow indicating the path from the icon representing the patient's current location to the target point is also drawn. The navigation terminal obtains its real-time location through a built-in indoor positioning module and compares it with the planned path. If it detects that the patient has deviated from the planned path beyond the allowable error range, it provides a correction prompt via voice or vibration. For example, if the allowable error range is set to 5 meters, the indoor positioning module uses Bluetooth beacons to locate the real-time coordinates returned by the navigation terminal. The navigation terminal calculates the shortest vertical distance from this location to the planned path. If the distance exceeds 5 meters, it triggers a voice prompt "You have deviated from the route, please turn left at the next intersection," while the device vibrates briefly. It is understood that the indoor positioning module can be implemented using ultra-wideband technology based on Wi-Fi fingerprinting, Bluetooth beacons, or UWB. Once a patient reaches a task point and completes the corresponding task, the navigation terminal receives confirmation of task completion via either an interactive interface or automatic sensing, and then sends this status back to the navigation management server. For example, upon reaching the "Pharmacy Window 3" area, a graphical interface pops up a "Confirm medication pickup complete?" button. After the patient clicks this button, the navigation terminal sends a task completion status update message to the navigation management server via the network. This message includes the task point identifier and a completion timestamp. Optionally, the automatic sensing method can detect when the patient has remained within the geofence of the task point for more than a preset time (e.g., 2 minutes) and automatically confirm task completion. Upon receiving the feedback, the navigation management server updates the patient's task execution status in the background and can provide a basis for subsequent data analysis.

[0043] In one embodiment of the present invention, during the guided execution process, the navigation management server continuously monitors specific environmental dynamic data change events reported by the IoT gateway. Event monitoring rules are set in the navigation management server, including: events where the number of people queuing at a specific treatment node exceeds a preset threshold, events where the passage speed of a key channel is lower than a preset threshold, and events where medical equipment changes from available to faulty. The navigation management server subscribes to specific topic messages published by the IoT gateway, continuously receives environmental dynamic data streams, performs real-time parsing of the received environmental dynamic data streams, and matches them with the event monitoring rules. When the real-time parsed data meets the conditions of any event monitoring rule, it is determined that an environmental dynamic data change event has been detected, and the type, location, and time of the corresponding event are recorded. When the detected event meets a preset replanning trigger condition, a new navigation planning scheme is regenerated based on the current real-time location of the patient and switched. The replanning trigger condition is predefined as follows: the detected event causes the increase in the path movement cost to the corresponding target node in the current navigation planning scheme to exceed the tolerance limit. Upon detecting changes in environmental dynamic data, the impact of the event on the currently executing navigation planning scheme is assessed, and the estimated waiting time increment for affected nodes or the movement cost increment for affected paths is calculated. The calculated time increment is compared with a preset tolerance limit; if the time increment exceeds the tolerance limit, the replanning trigger condition is met. Real-time location information reported by the patient's navigation terminal is obtained and used as the starting point for the new path planning. Based on the latest environmental dynamic data, the dynamic cost map is reconstructed, and an improved navigation planning algorithm is invoked to generate a new navigation planning scheme with the new starting point and the remaining unfinished task list. The new navigation planning scheme is sent to the patient's navigation terminal, which interrupts the current guidance and switches to execute the new navigation planning scheme.

[0044] In practical implementation, during the guided execution process, the navigation management server continuously monitors specific environmental dynamic data change events reported by the IoT gateway. Event monitoring rules are set in the navigation management server, including events such as: the number of people queuing at specific treatment nodes exceeding a preset threshold, the passage speed in key channels falling below a preset threshold, and medical equipment changing from available to faulty. For example, a queueing threshold of 15 people is set for the "ECG room" node, a passage speed threshold of 0.5 meters per second is set for the "outpatient main corridor" channel, and a monitoring event for "CT scan device 1" showing a "faulty" status is set. Data comparison shows that the monitoring conditions differ for different events. The navigation management server subscribes to specific topic messages published by the IoT gateway and continuously receives environmental dynamic data streams. Subscribed topics include "sensor / video / queue," "sensor / passage / traffic," and "sensor / device / status." The navigation management server parses the received environmental dynamic data stream in real time and matches it with event monitoring rules. When the parsed data meets the conditions of any event monitoring rule, it determines that an environmental dynamic data change event has been detected and records the corresponding event type, location, and time. For example, the navigation management server parses a message from the topic "sensor / video / queue" with the content {"node_id":"ecg_room","queue_num":20,"timestamp":"10:05:00"}. Matching it with the event monitoring rules, it finds that the number of people queuing in the "ECG room" (20) exceeds the preset threshold of 15. It then determines that a "queue number exceeding the preset threshold event for a specific treatment node" has occurred, records the event type as "queue exceeding limit", the location as "ECG room", and the time as "10:05:00".

[0045] In some embodiments, when a detected event meets a preset replanning trigger condition, a new navigation planning scheme is regenerated and switched based on the current real-time location of the patient. The replanning trigger condition is predefined as follows: the detected event causes the increase in the path movement cost to the corresponding target node in the current navigation planning scheme to exceed the tolerance limit. The tolerance limit is set in seconds; for example, the tolerance limit is set to 600 seconds for node waiting time increments and 300 seconds for path movement cost increments. After detecting a dynamic change event in environmental data, the impact of the event on the currently executing navigation planning scheme is evaluated, and the expected waiting time increment of the affected node or the movement cost increment of the affected path is calculated. The formula for calculating the time increment is expressed as: in: Indicates the time increment caused by the event. This represents the new time cost calculated based on the latest environmental dynamic data after the event (which may be the node's estimated waiting time or the path's predicted travel time). This represents the old time cost used in the original navigation planning scheme. For example, for a monitored event of "excessive queue length in the ECG room," the new estimated waiting time for the ECG room is calculated based on the latest environmental data. The estimated waiting time is 1200 seconds, while the original plan used a different estimated waiting time. If the time interval is 300 seconds, then the node waiting time increment is... Seconds. For example, for the event of "low travel speed in the main outpatient corridor," a new predicted travel time through that corridor is calculated. The original plan used a travel time of 300 seconds. If the time is 150 seconds, then the incremental cost of path movement is... Second.

[0046] The process involves comparing the calculated time increment with a preset tolerance limit. If the time increment exceeds the tolerance limit, the replanning trigger condition is met. The node waiting time increment of 900 seconds exceeds the node waiting time tolerance limit of 600 seconds, thus meeting the replanning trigger condition. The path movement cost increment of 150 seconds does not exceed the path movement cost tolerance limit of 300 seconds, therefore replanning is not triggered. The system obtains the real-time location information reported by the patient's navigation terminal and uses this information as the starting point for the new path planning. For example, if the navigation terminal reports its current coordinates as "floor 1, coordinates (X=105, Y=200)" through indoor positioning, the navigation management server maps these coordinates to the nearest vertex in the dynamic cost map, such as "center of the outpatient hall," and uses this vertex as the new path planning starting point. Based on the latest environmental dynamic data, the navigation management server reconstructs the dynamic cost map and calls an improved navigation planning algorithm to generate a new navigation planning scheme with the new starting point and the remaining unfinished task list. In practice, the remaining unfinished task list is obtained by removing the task points that the navigation terminal has confirmed as completed from the original total task sequence. For example, if the original task sequence is ["Pharmacy Window 3", "ECG Room", "CT Room"], and the navigation terminal has confirmed that the task "Pharmacy Window 3" is completed, then the remaining task list is ["ECG Room", "CT Room"].

[0047] In practice, the dynamic cost map is reconstructed based on the latest environmental dynamic data. The process is the same as the initial reconstruction, but uses the latest sensor data after the event occurred. For example, the queue length for the "ECG room" has been updated to 20 people, and its dwell cost is recalculated; the passage speed in the "outpatient main corridor" has been updated, and its movement cost is recalculated. An improved navigation planning algorithm is invoked, starting from the "central part of the outpatient hall" and using ["ECG room", "CT room"] as the target task points. Path planning is performed on the new dynamic cost map to generate a new task sequence, such as ["CT room", "ECG room"], to avoid the current high congestion in the "ECG room". In some embodiments, the replanning process is executed asynchronously and quickly in the background of the navigation management server, with the total time controlled within a few seconds to ensure the continuity of navigation guidance. The new navigation plan is then sent to the patient's navigation terminal, which interrupts the current guidance and switches to execute the new navigation plan. For example, if a navigation terminal is initially guiding a patient to the "ECG room," upon receiving a new navigation plan, it immediately interrupts the current arrow guidance and voice prompts, displays a message on the graphical interface stating "Route updated," and begins navigation according to the new sequence ["CT room," "ECG room"]. It recalculates and displays the path from the current location "Center of the Outpatient Hall" to the new first destination, the "CT room." Optionally, when switching plans, the navigation terminal may announce via voice that "Due to a long queue ahead, your route has been replanned; please proceed to the CT room first." The navigation management server records this replanning event log, including the triggering reason, the original plan, the new plan, and the triggering time, for subsequent system analysis and optimization.

[0048] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A hospital visit navigation method based on Internet of Things (IoT) technology, characterized in that, Includes the following steps: The system collects patient identification information by deploying an identification unit on the patient's medical card or mobile terminal and sends it to the navigation management server. Based on the received patient identification information, the navigation management server retrieves the patient's medical task list from the hospital information system. The medical task list includes several tasks to be performed and their associated physical location information. The navigation management server aggregates dynamic environmental data collected in real time from various sensors deployed within the hospital based on an IoT gateway. This dynamic environmental data includes the number of people queuing at each treatment node, the availability of equipment, and the real-time passage status of key channels. The navigation management server constructs a dynamic cost graph containing path travel time and node waiting time based on the dynamic environmental data, and calls an improved navigation planning algorithm to generate a navigation planning scheme containing task sequences and execution time windows based on the dynamic cost graph and the medical visit task list; the improved navigation planning algorithm is an improved ant colony optimization algorithm. The generated navigation plan is sent to the patient's navigation terminal to guide the patient in its execution.

2. The hospital visit navigation method based on Internet of Things (IoT) technology according to claim 1, characterized in that, The navigation management server aggregates dynamic environmental data collected in real time from various sensors deployed within the hospital based on an IoT gateway, specifically including: Video analytics sensors are deployed in front of clinics, examination rooms, and pharmacies. These sensors collect and analyze video streams in real time, count the number of people entering specific areas, and output the current number of people in queue. Integrate operational status sensors within medical devices to collect real-time data on device power-on / off status, current operating mode, and estimated idle time, and output device availability status data. Infrared beam sensors or lidar sensors are deployed in building corridors, elevator entrances, and stairwells to collect real-time data on the flow and speed of pedestrians or vehicles passing through specific channel sections, and output real-time traffic status data of key channels. Through the IoT gateway, a unified message queue telemetry transmission protocol is used to asynchronously receive reported data from video analytics sensors, operational status sensors, infrared beam sensors, and lidar sensors. The data reported by various sensors received through the IoT gateway are parsed for format, timestamp aligned, and outlier cleaned to form a structured and time-synchronized dynamic environmental dataset.

3. The hospital visit navigation method based on Internet of Things technology according to claim 2, characterized in that, The navigation management server constructs a dynamic cost map containing path travel time and node waiting time based on the dynamic environmental data, specifically including: Using the path intersections and treatment nodes on the hospital's indoor electronic map as vertices and the connecting paths between vertices as edges, an initial topology graph is constructed. Extract real-time queue number data and equipment availability status data for each treatment node from the environmental dynamic dataset; Based on the queuing theory model, real-time queuing data and equipment availability data are converted into the patient's expected waiting time at the corresponding treatment node, and the expected waiting time is used as the dwell cost of the corresponding node vertex. Real-time traffic status data of each key channel is extracted from the environmental dynamic dataset. Combined with the static physical length of the channel, the predicted travel time of pedestrians through the corresponding path segment is calculated, and the predicted travel time is used as the movement cost of the corresponding edge. The dwell cost and movement cost are respectively assigned to the vertices and edges of the initial topology graph to generate a dynamic cost graph that includes dynamic time costs.

4. The hospital visit navigation method based on Internet of Things technology according to claim 3, characterized in that, The queuing theory model converts real-time queuing data and equipment availability data into the patient's estimated waiting time at the corresponding treatment node, specifically: Configure the corresponding number of service desks and service intensity parameters for different types of diagnosis and treatment nodes; Use the real-time queue length data acquired by the sensor as the input to the queuing theory model; Convert the available status data of the equipment into the number of valid service desks. If the equipment status is faulty or occupied, subtract the corresponding number from the total number of service desks. By substituting the current queue length, number of effective service counters, and service intensity parameters into the calculation formula in the queuing theory model, the patient's expected waiting time can be calculated.

5. The hospital visit navigation method based on Internet of Things (IoT) technology according to claim 1, characterized in that, The improved navigation planning algorithm is invoked to generate a navigation planning scheme containing task sequences and execution time windows based on the dynamic cost map and the medical visit task list, including: The patient's starting point, each task point in the medical task list, and the ending point are mapped to the corresponding vertices in the dynamic cost graph, which serve as the target nodes that the algorithm needs to access. Initialize the pheromone matrix. The value of each element in the pheromone matrix represents the virtual pheromone concentration of the path connecting two target nodes. The initial concentration is set according to the reciprocal of the sum of the predicted travel time of the corresponding path and the expected waiting time of the target node in the dynamic cost map. Release multiple virtual ants. Each virtual ant starts from the starting point and selects the next target node to visit according to the state transition rules. The state transition rules comprehensively consider the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node. Once a virtual ant has completed the path construction from the starting point, visited all task points, and finally reached the destination, the total dynamic cost of the path is calculated. The total dynamic cost is the sum of the movement costs of all nodes on the path and the dwell costs of all visited nodes. Based on the total dynamic cost of the virtual ant's path construction, the corresponding values ​​of the pheromone in the pheromone matrix are updated according to the rule that the lower the cost, the more pheromone is released. After multiple virtual ants complete the iteration, the path with the lowest total dynamic cost in the historical iterations is selected as the optimal solution output. The resulting path is the navigation planning scheme containing the task sequence. Based on the task sequence in the navigation planning scheme, and combined with the predicted travel time between each task point and the estimated waiting time at each node, the estimated time for the patient to arrive at each task point is calculated, forming an execution time window.

6. The hospital visit navigation method based on Internet of Things technology according to claim 5, characterized in that, The state transition rule comprehensively considers the pheromone concentration on the candidate node path, the real-time movement cost of the corresponding path in the dynamic cost graph, and the real-time dwell cost of the next target node, specifically: Calculate the transition probability from the current node to each unvisited candidate node. The transition probability is positively correlated with the pheromone concentration on the path connecting the current node and the candidate node, positively correlated with the inverse of the real-time movement cost of the corresponding path in the dynamic cost graph, and positively correlated with the inverse of the real-time residence cost of the candidate node in the dynamic cost graph. Based on the calculated transition probabilities of each candidate node, the roulette wheel method is used to select the next node to be actually visited.

7. A hospital visit navigation method based on Internet of Things (IoT) technology according to claim 5, characterized in that, After multiple virtual ants complete the iterations and select the path with the lowest total dynamic cost from the historical iterations as the optimal solution, the process further includes verifying the time feasibility of the navigation planning scheme, specifically: Extract the task sequence and the predicted travel time between each task point from the navigation planning scheme, as well as the estimated waiting time at each task point; Obtain the appointment time limits or latest completion time requirements associated with each medical task from the hospital information system; Based on the task sequence, starting from the current system time, accumulate the travel time and waiting time to calculate the estimated arrival time and estimated departure time of each task point; The estimated departure time for each task point is compared with the latest completion time requirement for the corresponding task obtained from the hospital information system; If the estimated departure time of any task point is later than its latest completion time requirement, the current navigation planning scheme is deemed infeasible, and the task sequence needs to be readjusted or an alarm needs to be triggered. If the estimated departure time of all task points is no later than their latest completion time requirement, the navigation planning scheme is deemed feasible and the output is confirmed.

8. A hospital visit navigation method based on Internet of Things (IoT) technology according to claim 1, characterized in that, The step of sending the generated navigation plan to the patient's navigation terminal and guiding the patient to execute it specifically includes: The navigation management server encodes the navigation planning scheme, which includes task sequences and execution time windows, into a specific data protocol format; The encoded navigation plan is sent to the patient's navigation terminal via wireless local area network; The navigation terminal decodes the received navigation plan and displays the current task, the next task target point, and the arrow guide path from the current location to the next target point on the graphical interface; The navigation terminal obtains its real-time location through the built-in indoor positioning module and compares it with the planned path. If it detects that the patient deviates from the planned path beyond the allowable error range, it will provide correction prompts through voice or vibration. Once the patient reaches a task point and completes the corresponding task, the navigation terminal receives confirmation of task completion through interface interaction or automatic sensing, and sends this status back to the navigation management server.

9. A hospital visit navigation method based on Internet of Things (IoT) technology according to claim 1, characterized in that, The method further includes: During the guided execution process, the system continuously monitors specific environmental dynamic data change events reported by the IoT gateway. When the monitored events meet the preset replanning trigger conditions, a new navigation planning scheme is regenerated and switched based on the current real-time location of the patient. During the boot process, continuous monitoring of specific environmental dynamic data change events reported by the IoT gateway includes: Set event monitoring rules in the navigation management server. The event monitoring rules include: events such as the number of people queuing at a specific treatment node exceeding a preset threshold, events such as the passage speed of a key channel being lower than a preset threshold, and events such as medical equipment changing from an available state to a faulty state. The navigation management server subscribes to specific topic messages published by the IoT gateway and continuously receives dynamic environmental data streams. The received environmental dynamic data stream is parsed in real time and matched with the event monitoring rules; When the data parsed in real time meets the conditions of any event monitoring rule, it is determined that an environmental dynamic data change event has been detected, and the type, location and time of the corresponding event are recorded.

10. A hospital visit navigation method based on Internet of Things (IoT) technology according to claim 9, characterized in that, When the monitored event meets the preset replanning trigger condition, a new navigation planning scheme is regenerated and switched based on the current real-time location of the patient, specifically including: A replanning trigger condition is predefined, wherein the replanning trigger condition is: the detected event causes the increase in the path movement cost to the corresponding target node in the current navigation planning scheme to exceed the tolerance limit; After detecting environmental dynamic data change events, assess the impact of the event on the currently executing navigation planning scheme, and calculate the expected waiting time increment of affected nodes or the movement cost increment of affected paths; The calculated time increment is compared with the preset tolerance limit. If the time increment exceeds the tolerance limit, the replanning trigger condition is determined to be met. Obtain the real-time location information reported by the patient's navigation terminal, and use the real-time location information as the starting point for new path planning; Based on the latest environmental dynamic data, the dynamic cost map is reconstructed, and an improved navigation planning algorithm is invoked to generate a new navigation planning scheme with a new starting point and a list of remaining unfinished tasks. The new navigation plan is sent to the patient's navigation terminal, which then interrupts the current guidance and switches to execute the new navigation plan.

Images