A multi-objective path optimization method and system for emergency transfer

Abstract

The application belongs to the technical field of patient transfer, and discloses a multi-target path optimization method and system for emergency referral. The method comprises the following steps: acquiring multi-source data, pre-processing the multi-source data to obtain standardized data corresponding to each data category; performing fusion processing on the standardized data of each category to obtain feature data; based on the feature data, performing multi-target route optimization calculation to obtain an optimal referral route and at least one alternative referral route; in the referral process, monitoring the changes of the feature data and / or the multi-source data, and triggering route re-planning to update the optimal referral route and the at least one alternative referral route when a preset trigger condition is met. Through the deep fusion of GPS and real-time traffic data and in combination with the exclusive needs of emergency referral, the application effectively solves the defects of low data fusion precision, non-clinical demand planning and dynamic adjustment lag in the prior art, and significantly improves the accuracy, dynamic nature and practicality of the emergency referral route planning.

Description

Technical Field

[0001] This invention belongs to the field of patient transport technology, and specifically discloses a multi-objective path optimization method and system for emergency referral. Background Technology

[0002] As urban healthcare systems continue to improve, the connection between pre-hospital emergency care and in-hospital treatment becomes increasingly important. When ambulances are on transfer missions, they need to deliver patients to medical institutions with the appropriate treatment capabilities within a limited time; therefore, route selection directly affects the treatment outcome. Currently, common transfer route planning relies heavily on navigation systems to calculate based on real-time traffic conditions, usually using the shortest time or shortest distance as the primary criteria. While this method meets the needs of general travel scenarios, it has significant shortcomings in emergency transfer scenarios.

[0003] Current route planning primarily relies on vehicle location information and traffic data. While this reflects road conditions, it lacks consideration for the patient's condition. Different patients have varying requirements for transport timeliness and smoothness; a route selection solely based on time cannot adequately address these factors. Furthermore, existing systems typically do not incorporate hospital capacity into route decision-making. In actual referrals, the availability of beds, equipment, and specialist doctors at different hospitals is constantly changing. Even if an ambulance arrives at the target hospital quickly, it may face situations where timely admission is not possible, thus delaying treatment. Therefore, when external conditions change significantly, the inability to adjust routes promptly impacts referral efficiency and safety. Summary of the Invention

[0004] In view of this, the present invention provides a multi-objective path optimization method and system for emergency referral to solve the above problems.

[0005] The specific implementation of this invention is as follows: A multi-objective path optimization method for emergency referral includes: Acquire multi-source data, including: GPS location data of ambulances, real-time traffic data, medical condition data of patients awaiting transfer, and resource data of at least one target hospital; Preprocessing of multi-source data yields standardized data corresponding to each data category; Standardized data from various categories are fused and processed to obtain characteristic data representing the referral decision status. Based on feature data, multi-objective route optimization calculations are performed to obtain the optimal transfer route from the current location of the ambulance to the selected target hospital, as well as at least one alternative transfer route. During the referral process, monitor changes in characteristic data and / or multi-source data, and trigger route replanning to update the optimal referral route and at least one alternative referral route when preset trigger conditions are met; Output the optimal referral route and at least one alternative referral route to the associated terminal.

[0006] As an optional approach, multi-source data can be preprocessed to obtain standardized data corresponding to each data category, including: For GPS positioning data, a filtering algorithm is used to correct the positioning error, and abnormal positioning data points are removed based on statistical rules. For real-time traffic data, outliers are identified and removed, and for road segments with missing data, interpolation is performed to complete the data based on traffic data from adjacent road segments.

[0007] As an optional approach, preprocessing multi-source data to obtain standardized data corresponding to each data category also includes: Based on the patient's severity level in the medical data, a corresponding urgency score is generated; Based on the patient's positional protection needs in the disease data, a corresponding safety factor is generated; Based on the bed availability, equipment availability, and specialist physician on-duty status in the resource data, a corresponding diagnosis and treatment matching score is generated.

[0008] As an optional approach, standardized data from various categories are fused to obtain characteristic data representing the referral decision status, including: Based on the importance scores of standardized data for each category, the attention weight of each category of data in the fusion process is dynamically calculated; The importance score is determined based on at least one of the following: The positioning accuracy of GPS positioning data, the credibility of real-time traffic data sources, the urgency of patients reflected in disease data, and the matching degree of hospital diagnosis and treatment reflected in resource data; The calculated attention weights are used to weight and combine the feature vectors corresponding to the standardized data of each category to generate feature data.

[0009] As an optional approach, the determination of importance scores includes: assigning corresponding credibility scores as the importance scores of real-time traffic data based on the category of the real-time traffic data source, with the credibility scores of data from traffic management platforms being higher than those from map APIs; Furthermore, when real-time traffic data from different sources conflict, a weighted decision is made based on the credibility score of each source, and historical traffic data from the same period is used for auxiliary verification to resolve the conflict.

[0010] As an optional approach, multi-objective route optimization calculations can be performed based on feature data, including: Construct a multi-objective optimization function to evaluate the overall advantages and disadvantages of the route. Its objectives include at least minimizing the total referral time, minimizing the transfer risk, and maximizing the matching degree of diagnosis and treatment at the target hospital. Based on the patient's urgency, road network conditions, and patient safety needs reflected in the feature data, the weight coefficients corresponding to each objective in the multi-objective optimization function are dynamically adjusted. An improved shortest path search algorithm is adopted, which uses a multi-objective optimization function with dynamic weights as the path cost evaluation standard to search for the optimal referral route and at least one alternative referral route in the road network.

[0011] As an alternative approach, factors for assessing transshipment risk include the level of bumpiness of the route and / or the complexity of the route.

[0012] As an optional method, the preset trigger conditions include at least one of the following: The congestion level on a certain section of the planned route worsens beyond the first threshold. The patient's vital signs fluctuated beyond the second threshold; The treatment matching score of the selected target hospital dropped below the third threshold. The treatment matching score was generated based on resource data. The error in the GPS positioning data exceeds the fourth threshold; where the error in the GPS positioning data is obtained by processing the GPS positioning data.

[0013] As an optional approach, information on the optimal referral route and at least one alternative referral route is output to the associated terminal, including: The navigation information, estimated arrival time, and patient condition summary of the optimal referral route and at least one alternative referral route will be simultaneously sent to the ambulance's onboard terminal, the emergency command center's monitoring terminal, and the receiving terminal of the selected target hospital. The patient's condition summary is generated based on the patient's condition data and includes the patient's severity level and key vital signs.

[0014] On the other hand, the present invention also provides an emergency referral route optimization system that integrates GPS and real-time traffic data, comprising: The data acquisition module is used to acquire data from multiple sources; The data preprocessing module is used to preprocess multi-source data to obtain standardized data corresponding to each data category; The data fusion module is used to fuse standardized data of various categories to obtain feature data that characterizes the referral decision status; The route optimization module is used to perform multi-objective route optimization calculations based on feature data to obtain the optimal referral route and at least one alternative referral route. The dynamic adjustment module is used to monitor changes and trigger route replanning during the referral process; The output module is used to output information about the optimal referral route and at least one alternative referral route to the associated terminal.

[0015] The beneficial effects of this invention are as follows: This invention, through the deep integration of GPS and real-time traffic data, combined with the specific needs of emergency referrals, effectively solves the shortcomings of existing technologies such as low data fusion accuracy, planning that does not meet clinical needs, and lagging dynamic adjustments, and significantly improves the accuracy, dynamism, and practicality of emergency referral route planning. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the multi-objective path optimization method for emergency referrals according to the present invention. Figure 2 This is a schematic diagram illustrating an application scenario of the multi-objective path optimization method for emergency referral of the present invention; Figure 3 This is a data preprocessing flowchart of the multi-objective path optimization method for emergency referral of the present invention; Figure 4 This is a flowchart illustrating the calculation process of the multi-objective path optimization method for emergency referral in this invention. Figure 5 This is a flowchart of the route replanning process for the multi-objective path optimization method for emergency referrals of this invention; Figure 6 This is a schematic diagram of the system structure of the multi-objective path optimization method for emergency referral of the present invention. Detailed Implementation To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been presented in the various embodiments of the present invention to enable the reader to better understand the present invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and various changes and modifications based on the following embodiments.

[0017] In emergency transfer scenarios, ambulances navigating urban and highway networks are often simultaneously affected by location drift, sudden traffic congestion, and changes in the target hospital's capacity. For example, if GPS signals momentarily shift in elevated or tunnel areas, the navigation system may continue along the original route even as the road ahead becomes congested due to an accident. Meanwhile, intensive care beds at the originally planned hospital may be temporarily occupied, preventing timely treatment even if the patient arrives in the shortest possible time. In this process, existing route planning typically involves only one-time calculations based on time, lacking a coordinated response to changes in patient condition and medical resources, easily leading to a disconnect between route selection and actual treatment needs.

[0018] This embodiment addresses the problems of fragmented multi-source information and delayed path decision-making during the referral process by providing a multi-objective path optimization method for emergency referrals. By uniformly modeling and dynamically linking location information, traffic conditions, patient conditions, and hospital resources, the referral path can be adjusted in real time.

[0019] Please see Figure 1 First, multi-source data is acquired, including ambulance GPS positioning data, real-time traffic data, patient condition data, and resource data from at least one target hospital. Specifically, the ambulance periodically outputs its current location, speed, and direction information via its onboard positioning terminal, while simultaneously using the patient's personal device to obtain the patient's relative location, thus avoiding positional deviations during boarding / alighting or transfer. Real-time traffic data is obtained by accessing multiple data sources, with each source's data being tagged with its origin upon access for subsequent decision-making. Optionally, real-time traffic data includes road congestion indices, average vehicle speeds, and traffic incident information. This data originates from data interfaces opened by traffic management departments, government data platforms, traffic data interfaces provided by commercial map service platforms, and group perception data from mobile terminals (typically floating car data anonymously aggregated from the positioning information of a large number of vehicles), used to compensate for delays in traffic data updates. Furthermore, for traffic data from different sources, in addition to assigning credibility, a timeliness weight is introduced; the closer the data's generation time is to the current moment, the higher its weight in the fusion process, reducing the interference of outdated data on route decisions.

[0020] Patient condition data, collected in real-time by the vehicle-mounted emergency medical equipment, includes vital signs and severity level, indicating the urgency of the current transfer. Target hospital resource data is updated periodically via an interface, reflecting the capacity of each candidate hospital. This hospital resource data includes bed availability, equipment availability, and specialist physician on-duty status. Specifically, by statistically analyzing the ratio between the number of emergency room visits per unit time and the currently available resources, the hospital's current patient load is determined, reflecting the hospital's capacity to continue receiving patients within a short period. A scenario illustration is shown below. Figure 2 As shown.

[0021] In this embodiment, all types of data are accompanied by timestamps and aligned using a unified clock, so that data from different frequencies are comparable at the same time.

[0022] After acquiring the aforementioned multi-source data, the data is preprocessed to obtain standardized data corresponding to each data category. For details, please refer to [link to relevant documentation]. Figure 3 For GPS positioning data, which is prone to skipping or drifting in complex environments, this embodiment introduces a filtering mechanism to smooth the positioning results across consecutive time points. It also uses statistical rules to identify and remove outliers that deviate from the normal trajectory, ensuring the continuity of the positioning trajectory. For real-time traffic data, which varies in granularity and update cycle across different sources, this embodiment uses unified encoding and completes missing road segment data to ensure spatial continuity of the road network. For patient condition data, raw vital signs and clinical grades are converted into quantifiable indicators for subsequent calculations. For hospital resource data, bed availability, equipment, and personnel status are converted into unified, comparable indicators, and a set of hospitals in an available state is selected. Through these processes, the original heterogeneous data becomes more consistent in structure and dimensions.

[0023] After data standardization, the standardized data of each category are fused to obtain feature data representing the referral decision status. Specifically, this embodiment assigns corresponding weight parameters to different data based on their role in the current decision. Data with higher positioning accuracy has a larger weight in path determination, data with higher urgency of the condition has a higher weight, and resource data has a stronger impact on the decision when hospital resources are strained. Based on this, each type of data is mapped into a feature vector and combined according to the weights to form a unified decision feature expression. When data from different sources conflict, such as inconsistent road conditions, this embodiment prioritizes data with higher reliability and verifies it in conjunction with historical similar cases to reduce the probability of misjudgment.

[0024] Based on the aforementioned feature data, multi-objective route optimization calculations are performed to obtain the optimal transfer route from the ambulance's current location to the selected target hospital, as well as at least one alternative transfer route. This embodiment constructs a comprehensive evaluation function that includes time cost, transfer risk, and treatment matching degree. Time cost reflects travel time, transfer risk is related to road surface smoothness and path complexity, and treatment matching degree measures the target hospital's suitability for the current patient. The weights of these indicators are not fixed under different transfer scenarios. For example, when the severity of the illness is high, the time weight is increased, while when there are positional restrictions, the safety-related weight is increased. Subsequently, using the comprehensive evaluation function as the path cost in the road network topology, feasible paths are searched, and the path with the best evaluation result is selected as the primary path. Paths with similar evaluation results are selected as alternative paths to cope with subsequent changes.

[0025] During the referral process, changes in the characteristic data and / or the multi-source data are monitored, and route replanning is triggered when preset trigger conditions are met to update the optimal referral route and the at least one alternative referral route. Specifically, this embodiment continuously tracks changes in traffic conditions, patient vital signs, and hospital resources. When key parameters are detected to exceed set ranges, such as a significant increase in congestion on the road ahead or a decrease in the target hospital's capacity, it is determined that the current route may no longer be applicable, and the fusion and optimization process is re-executed based on the latest data. During the replanning process, sections of the original route that remain effective are prioritized to avoid frequent and significant adjustments that could interfere with driving operations.

[0026] Finally, the information of the optimal referral route and at least one alternative referral route is output to the associated terminals. Specifically, the information includes route navigation guidance, estimated arrival time, and a summary of the patient's condition, and is simultaneously sent to the ambulance's onboard terminal, the emergency command center, and the target hospital, enabling all parties to make synchronized decisions based on the same information.

[0027] In this embodiment, multi-source data is acquired in parallel, simultaneously obtaining ambulance GPS location data, real-time traffic data from traffic management platforms and map APIs, patient condition data awaiting transfer, and resource data from at least one target hospital. This data is aligned using timestamp synchronization technology to ensure time consistency in subsequent processing.

[0028] In this embodiment, preprocessing of multi-source data is performed to transform heterogeneous, noisy raw data into standardized, usable data, providing high-quality input for subsequent fusion. First, GPS positioning data and real-time traffic data are preprocessed. For GPS positioning data, a Kalman filter algorithm is used to correct the positioning error. The formula is:

[0029] in The corrected GPS positioning coordinates (latitude and longitude) at time k. Here are the corrected positioning coordinates at time k-1; A is the state transition matrix (value 1, due to the linear change in positioning state); B is the control matrix (value 0, due to no external control interference). This is the control vector (with a value of 0). Set the Kalman gain to 0.8 (to balance positioning accuracy and response speed). The values ​​are the original GPS positioning coordinates at time k. The 3σ criterion is used to remove outlier data (data with a deviation from the mean greater than 3 times the standard deviation is considered outlier).

[0030] The Kalman filter algorithm can effectively smooth positioning drift caused by tunnels, tall buildings, etc., through iterative prediction and updating. Simultaneously, it removes outlier positioning data points based on the 3σ statistical rule; that is, when the deviation of the original positioning data from the mean exceeds three times the standard deviation, it is identified as an outlier and removed. For real-time traffic data, outliers (such as negative vehicle speeds or speeds far exceeding road speed limits) are first identified and removed. For road segments missing due to data blind spots or delays, linear interpolation is used to complete the data based on traffic data from adjacent road segments. The linear interpolation formula is:

[0031] in, Traffic data (congestion index / vehicle speed) for blind spot road sections. Traffic data for two adjacent segments of the blind spot; The road segment number is used to assign numbers to two adjacent road segments; Number the blind spot road section.

[0032] Through the above processing, GPS positioning data and real-time traffic data have continuity in both spatial and temporal dimensions.

[0033] Furthermore, the data preprocessing process in this embodiment also includes standardizing the patient's condition data and the target hospital's resource data so that they can participate in unified path optimization calculations.

[0034] Specifically, the target hospital resource data includes bed availability, equipment availability, and specialist physician on-duty status. In this embodiment, a treatment matching score M is generated based on the above resource data, with a maximum score of 100 points.

[0035] in The availability of beds is scored (≥3 vacant beds = 100 points, 1-2 vacant beds = 70 points, 0 vacant beds = 0 points). Rate the availability of equipment (all required equipment available = 100 points, partially available = 60 points, none available = 0 points). The on-duty score for specialist doctors is calculated (on-duty = 100 points, off-duty = 0 points).

[0036] In the specific processing, the number of available beds is graded and assigned a score based on the number of available beds. A score of 100 is assigned when there are at least 3 available beds, 70 points when there are 1 to 2 available beds, and 0 points when there are 0 available beds. Equipment availability is assigned a score based on the availability of the equipment required by the target patient. A score of 100 is assigned when all required equipment is available, 60 points when some equipment is available, and 0 points when none is available. Specialist physician on-duty status is assigned a score based on whether the corresponding specialist physician is on duty. A score of 100 is assigned when on duty, and 0 points when off duty.

[0037] Based on the above, the bed occupancy score, equipment availability score, and specialist physician on-duty score are combined and calculated to obtain the diagnosis and treatment matching score M, so that the reception capacity of different hospitals can be represented by a unified value.

[0038] In this embodiment, the feature data refers to a multi-dimensional feature set formed by unifying GPS positioning data, real-time traffic data, patient condition data, and target hospital resource data. This feature set comprehensively reflects the current ambulance location status, road network traffic status, patient condition status, and target hospital reception status. Compared to the original data, the feature data has undergone dimensional unification and structured processing, and can be directly used as input parameters for subsequent route optimization calculations.

[0039] Specifically, the feature data includes GPS positioning features, traffic features, patient condition features, and hospital resource features. GPS positioning features characterize the ambulance's current location, speed, and positioning accuracy; traffic features characterize the congestion level, traffic efficiency, and data reliability of the current road segment; patient condition features include the patient's urgency score S (Level I = 5 points, Level II = 3 points, Level III = 1 point) and the safety coefficient C corresponding to the need for postural protection (no special needs = 1.0, minor contraindications = 1.2, severe contraindications = 1.5), which can be expanded based on changes in vital signs; hospital resource features include the treatment matching score M (out of 100) and its related resource status parameters.

[0040] Considering the dynamic fluctuations in the contribution and reliability of each data source to the final route decision at different referral stages, this embodiment introduces an attention mechanism to dynamically allocate fusion weights for each category of data. Specifically, after obtaining the aforementioned features, the standardized data for each category is fused. First, attention weights are calculated based on the importance scores of the standardized data for each category. The importance score is determined by the real-time status of the data: for GPS positioning data, its score is inversely proportional to its positioning accuracy factor; for real-time traffic data, its score is linked to the credibility of its source; for disease status data, its score is determined by the aforementioned urgency score; and for resource data, its score is determined by the diagnosis-treatment matching score.

[0041] This embodiment uses these scores to calculate the attention weights for each category of data using an exponential normalization function. The calculation formula is as follows:

[0042] Where i = 1, 2, 3, 4 (corresponding to GPS, traffic, medical condition, and hospital resource data, respectively). Parameter description: Attention weights for the i-th class of data ( ); Score the importance of the i-th type of data (GPS data S1 = positioning accuracy / 10, traffic data S2 = data reliability, medical condition data S3 = urgency score / 5, hospital resource data S4 = diagnosis and treatment matching degree / 100).

[0043] Furthermore, after obtaining the attention weights for each category of data, the standardized data for each category are mapped to corresponding feature vectors, and then weighted and combined according to the attention weights to generate unified feature data, calculated as follows:

[0044] in, The fused feature vector; This is a GPS positioning feature vector; For traffic feature vectors; This represents the feature vector of the disease condition; These are hospital resource feature vectors, and the dimensions of each feature vector are determined based on the actual data fields collected.

[0045] During the fusion process, when conflicts arise between real-time traffic data from different sources (GPS indicates an ambulance is on a certain road segment, while traffic management data shows that the segment is closed), weights are allocated based on the credibility scores of each source, prioritizing data with higher credibility for feature construction. Simultaneously, historical traffic data from the same period is used to assist in verifying the conflict results, reducing the impact of abnormal data on the fusion outcome. Furthermore, among the traffic data sources, data from the traffic management platform has a higher credibility score than data from the map API, making the fusion result more closely reflect actual road conditions.

[0046] Furthermore, this embodiment performs multi-objective route optimization calculations based on feature data to obtain the optimal transfer route from the ambulance's current location to the selected target hospital, as well as at least one alternative transfer route, including: Please see Figure 4 A multi-objective optimization function is used to quantitatively evaluate the overall merits of candidate routes. This function includes at least three evaluation indicators: total transfer time, transfer risk, and target hospital treatment matching degree. The total transfer time reflects the estimated time it takes for the ambulance to travel from the current location to the target hospital; the transfer risk reflects the potential adverse effects on the patient during transfer; and the target hospital treatment matching degree reflects the hospital's suitability for the current patient. Based on this, the above evaluation indicators are weighted and combined to obtain the objective function value used for route evaluation.

[0047] in, To optimize the objective function value (the smaller the value, the better the route); These are time weight, safety weight, and treatment weight, respectively, and the sum of the three is 1; Total referral time; R represents the transshipment risk, which can be set to level 1-5, with level 1 being the lowest and level 5 being the highest. M represents the matching degree of the target hospital's diagnosis and treatment.

[0048] In the aforementioned multi-objective optimization function, the weight coefficients corresponding to each evaluation index are not fixed but are typically dynamically adjusted based on the real-time status reflected by the feature data. In some implementation scenarios, when the patient's urgency score S is high, the weight corresponding to the total transfer time is increased; when the safety factor C corresponding to the patient's postural protection needs is high, the weight corresponding to the transfer risk is increased; and when the treatment matching scores M of candidate hospitals differ significantly, the weight corresponding to the treatment matching degree is increased. During this process, the weight coefficients are calculated based on factors such as the patient's urgency score S, traffic congestion level, road network complexity, and safety factor C.

[0049] in, Rate the urgency of the patient's condition (1-5 points); The average congestion index of the route (level 0-10); The road network complexity is categorized into levels 1-3, with level 1 being the simplest. The patient's positional safety factor is 1.0-1.5. The route is classified as bumpy (levels 1-5, with level 1 being the smoothest).

[0050] Furthermore, the assessment of transfer risk is not only related to traffic conditions but also to the structural characteristics of the route itself. Factors influencing transfer risk assessment may include the route's bumpiness level and / or route complexity. Bumpiness level reflects the smoothness of the road surface and can be graded based on road type or historical driving data; route complexity reflects the number of turns, intersections, or the proportion of complex road sections in the route. A higher risk level corresponds to a greater number of consecutive turns or complex road sections. All of the above can be pre-set before implementing this embodiment. Based on this, the bumpiness level and route complexity are quantified and incorporated into the transfer risk calculation process.

[0051] Please see Figure 5 After constructing the multi-objective optimization function and dynamically adjusting the weights, this embodiment employs an improved shortest path search algorithm to solve for paths in the road network. Based on the traditional Dijkstra's algorithm, an optimization objective function Z is incorporated, using the Z-value of each route as the path weight. All feasible routes are traversed (with the ambulance's current location as the starting point and alternative hospitals as the destination), and the route with the smallest Z-value is selected as the optimal route. Two to three routes with a Z-value difference ≤ 10% from the optimal route are selected as alternative routes. This process generates the optimal route, alternative routes, and detailed information.

[0052] Meanwhile, to improve the fault tolerance of the path, paths whose objective function value differs from the optimal path within a preset range are selected as alternative referral routes from all candidate paths. In this embodiment, the preset range can be set according to actual needs, so that the alternative routes are close to the optimal path in overall performance, thereby enabling rapid switching during subsequent path adjustments.

[0053] Given the established route, the actual referral process is not a static execution process, but a dynamic one that continuously changes with the road network status, patient status, and hospital reception status. Therefore, this embodiment also includes a step to continuously monitor key data during the referral process and trigger route replanning when preset conditions are met.

[0054] In actual operation, the obtained feature data is the key monitoring object, while the original multi-source data is used for auxiliary judgment. Since the feature data has already fused and expressed multi-dimensional information, its changes can directly reflect the overall trend of the current referral environment, and therefore it is given priority as the trigger judgment basis.

[0055] Regarding the triggering mechanism, this embodiment sets quantization thresholds for different types of changes. As long as any condition is met, it is considered that the existing path is no longer suitable and the path needs to be recalculated.

[0056] In terms of traffic, when the congestion level on a certain section of the planned route deteriorates significantly within a short period—for example, if the congestion index rises by more than a preset range compared to the previous time window, or if a sudden accident or temporary traffic control occurs—it is determined that the traffic efficiency of that section has undergone a substantial change, thus triggering a replanning. This process can be based on the changing relationships of traffic parameters over consecutive time periods (optional). ).

[0057] In terms of patient status, vital signs are continuously monitored. When fluctuations in key indicators such as heart rate and blood pressure exceed preset ranges, or when the patient's severity level is assessed as increasing based on monitoring results, it indicates a change in the requirement for timely referral. At this point, the trigger path is recalculated. ).

[0058] In terms of hospital resources, when the treatment matching score M of the originally selected target hospital drops below the preset threshold, for example, due to bed occupancy or temporary equipment unavailability leading to a decrease in patient capacity, it indicates that the endpoint conditions have changed, and it is necessary to reselect the target hospital and update the corresponding path. ).

[0059] In addition, in terms of positioning, when the GPS positioning error exceeds the preset range, or when the positioning signal is interrupted, it will affect the path matching and navigation accuracy, and will also trigger path replanning.

[0060] Upon triggering any of the aforementioned conditions, this embodiment does not directly recalculate the entire path. Instead, it first performs local adjustments based on the current path execution progress. It prioritizes retaining road segments in the original path that still meet the current traffic conditions, replacing only the affected areas to avoid frequent large-scale detours that could disrupt driving. Simultaneously, during the replanning process, the aforementioned data fusion mechanism and multi-objective optimization method are continued to ensure the new path remains consistent with the current state. Furthermore, considering the potential for short-term data anomalies in real-world environments, this embodiment introduces a continuity constraint during trigger judgment. That is, replanning is only executed when a trigger condition is met within multiple consecutive sampling periods, reducing the likelihood of false triggers due to a single instance of abnormal data.

[0061] Considering the time constraints and limited fault tolerance in this scenario, the above solution, which only generates path results during the referral process, is insufficient to support synchronous collaboration among multiple parties. Therefore, it is necessary to organize path-related information and patient status information in a unified manner and send them synchronously to different terminals so that all participants can make decisions based on the same information.

[0062] Based on this, this embodiment combines the navigation information, estimated arrival time, and patient condition summary of the optimal referral route and at least one alternative referral route to form structured output data. The navigation information includes key road segments along the route and corresponding driving instructions; the estimated arrival time is calculated based on the current route and real-time traffic conditions; the patient condition summary is generated based on the condition data and includes at least the patient's severity level and key vital signs, enabling the receiving end to quickly grasp the patient's condition. The field settings and structure of this structured information are not limited in this embodiment.

[0063] During output, information is simultaneously sent to the ambulance's onboard terminal, the emergency command center's monitoring terminal, and the receiving terminal at the selected target hospital. The ambulance's onboard terminal primarily provides real-time navigation information and route update prompts, enabling drivers to navigate according to the current route. The emergency command center's monitoring terminal mainly displays the vehicle's location, route progress, and changes in the patient's condition for overall dispatching. The target hospital's receiving terminal primarily provides the estimated arrival time and a summary of the patient's condition to facilitate advance preparation for treatment.

[0064] Furthermore, during the referral process, the output information is updated synchronously as the route or patient status changes. Especially when route replanning is triggered, the updated optimal referral route and alternative route information are immediately resent to each terminal, ensuring that the information received by each terminal remains consistent with the current actual route. In addition, to reduce the complexity of data parsing between different terminals, the output data can be encapsulated using a unified data structure, ensuring that navigation information, estimated arrival time, and patient condition summaries maintain a consistent expression across different terminals. This embodiment does not impose any limitations on this and therefore will not elaborate further (in addition, necessary anonymization processing is performed on patient identity-related information during the output process, retaining only the necessary fields related to diagnosis and treatment, to balance information usability and data security requirements).

[0065] On the other hand, please see Figure 6 This embodiment also provides an emergency referral route optimization system that integrates GPS and real-time traffic data, characterized in that it includes: a data acquisition module for acquiring multi-source data; The data preprocessing module is used to preprocess multi-source data to obtain standardized data corresponding to each data category; the data fusion module is used to fuse standardized data of each category to obtain feature data representing the referral decision status; the route optimization module is used to perform multi-objective route optimization calculations based on feature data to obtain the optimal referral route and at least one alternative referral route; the dynamic adjustment module is used to monitor changes and trigger route replanning during the referral process; and the output module is used to output information about the optimal referral route and at least one alternative referral route to the associated terminal.

[0066] The embodiments of the present invention have been described in detail above. The various embodiments in the specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make several improvements and modifications to the present invention without departing from the principles of the invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims

1. A multi-objective path optimization method for emergency referral, characterized in that, include: Acquire multi-source data, including: GPS location data of ambulances, real-time traffic data, medical condition data of patients awaiting transfer, and resource data of at least one target hospital; The multi-source data is preprocessed to obtain standardized data corresponding to each data category; Standardized data from various categories are fused and processed to obtain characteristic data representing the referral decision status. Based on the aforementioned feature data, multi-objective route optimization calculations are performed to obtain the optimal transfer route from the current location of the ambulance to the selected target hospital, as well as at least one alternative transfer route. During the referral process, monitor changes in characteristic data and / or multi-source data, and trigger route replanning to update the optimal referral route and at least one alternative referral route when preset trigger conditions are met; Output the optimal referral route and at least one alternative referral route to the associated terminal.

2. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The preprocessing of multi-source data to obtain standardized data corresponding to each data category includes: For GPS positioning data, a filtering algorithm is used to correct the positioning error, and abnormal positioning data points are removed based on statistical rules. For real-time traffic data, outliers are identified and removed, and for road segments with missing data, interpolation is performed to complete the data based on traffic data from adjacent road segments.

3. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The preprocessing of multi-source data to obtain standardized data corresponding to each data category also includes: Based on the patient's severity level in the medical data, a corresponding urgency score is generated; Based on the patient's positional protection needs in the disease data, a corresponding safety factor is generated; Based on the bed availability, equipment availability, and specialist physician on-duty status in the resource data, a corresponding diagnosis and treatment matching score is generated.

4. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The process of fusing standardized data from various categories to obtain characteristic data representing the referral decision status includes: Based on the importance scores of standardized data for each category, the attention weight of each category of data in the fusion process is dynamically calculated; The importance score is determined based on at least one of the following: The positioning accuracy of GPS positioning data, the credibility of real-time traffic data sources, the urgency of patients reflected in disease data, and the matching degree of hospital diagnosis and treatment reflected in resource data; The calculated attention weights are used to weight and combine the feature vectors corresponding to the standardized data of each category to generate feature data.

5. The multi-objective path optimization method for emergency referral according to claim 4, characterized in that, The method for determining the importance score includes: assigning a corresponding credibility score as the importance score of the real-time traffic data based on the category of the real-time traffic data source, and the credibility score of the data sourced from the traffic management platform is higher than that of the data sourced from the map API. Furthermore, when real-time traffic data from different sources conflict, a weighted decision is made based on the credibility score of each source, and historical traffic data from the same period is used for auxiliary verification to resolve the conflict.

6. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The multi-objective route optimization calculation based on feature data includes: A multi-objective optimization function is constructed to evaluate the overall advantages and disadvantages of the route. Its objectives include at least minimizing the total referral time, minimizing the transfer risk, and maximizing the matching degree of diagnosis and treatment at the target hospital. Based on the patient's urgency, road network conditions, and patient safety needs reflected in the feature data, the weight coefficients corresponding to each objective in the multi-objective optimization function are dynamically adjusted. An improved shortest path search algorithm is adopted, which uses a multi-objective optimization function with dynamic weights as the path cost evaluation standard to search for the optimal referral route and at least one alternative referral route in the road network.

7. The multi-objective path optimization method for emergency referral according to claim 6, characterized in that, The assessment factors for the transshipment risk include the level of bumpiness of the route and / or the complexity of the route.

8. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The preset triggering condition includes at least one of the following: The congestion level on a certain section of the planned route worsens beyond the first threshold. The patient's vital signs fluctuated beyond the second threshold; The treatment matching score of the selected target hospital dropped below the third threshold. The treatment matching score was generated based on resource data. The error in the GPS positioning data exceeds the fourth threshold; where the error in the GPS positioning data is obtained by processing the GPS positioning data.

9. The multi-objective path optimization method for emergency referral according to claim 1, characterized in that, The step of outputting the optimal referral route and at least one alternative referral route to the associated terminal includes: The navigation information, estimated arrival time, and patient condition summary of the optimal referral route and at least one alternative referral route will be simultaneously sent to the ambulance's onboard terminal, the emergency command center's monitoring terminal, and the receiving terminal of the selected target hospital. The patient's condition summary is generated based on the patient's condition data and includes the patient's severity level and key vital signs.

10. An emergency referral route optimization system integrating GPS and real-time traffic data, characterized in that, include: The data acquisition module is used to acquire data from multiple sources; The data preprocessing module is used to preprocess multi-source data to obtain standardized data corresponding to each data category; The data fusion module is used to fuse standardized data of various categories to obtain feature data that characterizes the referral decision status; The route optimization module is used to perform multi-objective route optimization calculations based on feature data to obtain the optimal referral route and at least one alternative referral route. The dynamic adjustment module is used to monitor changes and trigger route replanning during the referral process; The output module is used to output information about the optimal referral route and at least one alternative referral route to the associated terminal.

Images