Multi-mode collaborative medical emergency resource scheduling system

Abstract

The invention provides a multi-mode cooperative medical emergency resource scheduling system. The system comprises a resource information integration module, an emergency demand sensing module, a multi-mode cooperative scheduling engine and a rescue process cooperative monitoring module. Collection and fusion of various data sources are realized through the resource information integration module, intelligent analysis of resource demands is realized by combining the emergency demand sensing module and utilizing big data analysis and machine learning technologies, task tracking feedback and resource use condition tracking are supported, and the accuracy and timeliness of resource scheduling are ensured; and meanwhile, through a rescue process cooperative monitoring module, safe and visual management of the whole emergency disposal process is realized.

Description

Technical Field

[0001] This invention relates to the field of emergency response to medical resources, and in particular to a multimodal collaborative medical emergency resource dispatch system. Background Technology

[0002] Existing hospital emergency resource dispatch systems can manage and dispatch resources through information technology. However, in public health emergencies, traditional systems rely on fragmented communication methods (such as telephone and SMS), resulting in delays of more than 3 minutes in the transmission of critical information (such as patient vital signs and on-site video) between ambulances and hospitals, thus delaying the "golden hour" of treatment. At the same time, the manual scheduling method is still used, resulting in the emergency department operating beyond its capacity (working more than 12 hours a day) while grassroots resources are idle (equipment idle rate of 40%). Moreover, the allocation of materials relies on paper ledger updates, which leads to a high rate of material mismatch under emergency needs and easily results in material shortages or stockpiles.

[0003] In addition, the existing system relies on manual screening of massive amounts of testing data (such as information sources for 11 types of infectious diseases), resulting in delayed situational awareness and long deployment times for epidemiological investigations during sudden outbreaks; moreover, the emergency response plan is fixed and cannot generate the optimal prevention and control plan in real time based on the transmission trajectory.

[0004] While some systems have attempted to introduce single-technology improvements (such as RFID for material tracking), they have not solved the fundamental problem of multimodal resource collaborative scheduling: heterogeneous resources such as ambulances, drones, and volunteers cannot be dynamically coordinated on a unified platform; and there are gaps in the transmission of instructions between medical, transportation, and communication departments. In other words, current medical emergency dispatch systems still suffer from a lack of multimodal collaborative capabilities, rigid resource scheduling mechanisms, and insufficient intelligent decision-making. Therefore, there is an urgent need to build an integrated dispatch system that supports data fusion, intelligent prediction, and multi-entity collaboration. Summary of the Invention

[0005] To address the above problems, this invention proposes a multimodal collaborative medical emergency resource scheduling system.

[0006] The main contents of this invention include: A multimodal collaborative medical emergency resource scheduling system includes: The resource information integration module is used to integrate the HIS system of medical institutions, GPS positioning of ambulances, population flow heat maps and historical epidemic data through federated learning technology to build a digital twin of medical resources and collect dynamic status data of ambulances, drone medical units, mobile emergency rescue units and volunteer resources in real time. The emergency demand perception module is used to quantify the priority of medical resource demand based on the type of emergency event and a three-tier assessment system that combines event severity, resource gap, and social impact. The multimodal collaborative scheduling engine connects the resource information integration module and the emergency demand perception module. It uses the NSGA-II algorithm to generate a multi-objective optimization scheme that aims to minimize resource transportation time and maximize the coverage rate of severe cases. It outputs collaborative scheduling instructions that include ambulance route planning, UAV airspace routes and volunteer task allocation. The rescue process collaborative monitoring module integrates a GIS geographic information system and a real-time video communication unit to dynamically display the location trajectory of rescue resources, injury heat maps, and compare and mark the golden 10 minutes for on-site first aid and the golden 1 hour for transfer. The contingency plan management module pre-sets three types of special contingency plans: natural disasters, accidents, and public health emergencies. Each type of contingency plan is associated with a specific resource scheduling strategy through predefined resource matching rules.

[0007] Preferably, the emergency demand perception module includes a transmission path tracking unit and an emergency room traffic prediction unit; the transmission path tracking unit is used to predict the risk of virus transmission in different regions, and the transmission risk is a risk level quantified based on the infectious disease transmission level and the number of potential infections; the emergency room traffic prediction unit is used to predict emergency room traffic based on influenza transmission areas and population flow data, and the emergency room traffic includes a resource gap level quantified based on the occupancy rate of beds and medical equipment and the number of critically ill patients.

[0008] Preferably, the transmission path tracking unit integrates operator roaming data, traffic checkpoint records, and public place monitoring, and uses a sequential convex programming algorithm to fuse traffic checkpoint data with meteorological parameters to construct a personnel movement trajectory map. Combined with real-time meteorological sensor data, it generates a virus transmission risk heat map based on the spatial diffusion characteristics of the Pacejka tire model.

[0009] Preferably, the emergency room traffic prediction unit integrates electronic medical records, drug sales, social media sentiment and meteorological data, and uses an improved SEIR spatiotemporal propagation model and an LSTM time series model, superimposed with the ARIMA algorithm to capture seasonal fluctuations, to predict the peak of outpatient visits in 72 hours. The prediction results are verified by the hospital HIS system bed occupancy rate and medical equipment utilization rate.

[0010] Preferably, a three-tiered assessment system is used to quantify the priority of medical resource needs based on the type of emergency, combined with the severity of the event, resource gaps, and social impact, including: Based on the heat map of virus transmission risk, areas that meet the set risk level are marked as areas awaiting rescue, and priority rescue areas corresponding to the areas awaiting rescue are determined. Obtain 72-hour peak medical visit forecast information for areas awaiting rescue and corresponding priority rescue areas, and calculate the corresponding resource gap level; Based on the resource gap level of the corresponding region, the priority of medical resource demand is quantified in combination with social impact. The social impact includes corresponding indices quantified based on social media sentiment and traffic control measures.

[0011] Preferably, the priority rescue areas corresponding to the areas to be rescued include: Obtain information on medical institutions within the designated area to be rescued; Based on the established evaluation criteria, organizations that meet the criteria are marked as priority rescue areas; The evaluation indicators include the completion rate of treatment within the golden hour and the abnormal turnover rate of supplies.

[0012] Preferably, the multimodal collaborative scheduling engine includes a scheduling feedback optimization unit and a scheduling instruction execution unit; the scheduling feedback optimization unit is used to optimize and adjust the collaborative scheduling instructions based on the task execution status and historical events; the scheduling instruction execution unit is used to decompose the collaborative scheduling instructions of the multi-objective optimization scheme generated according to the priority of medical resource demand into executable instructions and send them to the corresponding execution terminals.

[0013] Preferably, the scheduling scheme optimization unit monitors resource consumption through IoT sensors, calculates the deviation from the predicted value, and dynamically triggers the correction of the scheduling scheme when the actual consumption differs from the predicted value by more than the predicted value, and updates the parameters of the demand prediction model through federated learning.

[0014] Preferably, the execution process of the scheduling instruction execution unit includes: Send navigation routes and target hospital information to the ambulance terminal; Send the coordinates of the material delivery and emergency obstacle avoidance instructions to the drone terminal; Push information on the assembly points for lightly injured patients and safety guidelines to the volunteer terminals.

[0015] Preferably, it also includes a collaborative communication module, which is used to achieve one-click distribution of task instructions based on voiceprint authentication through satellite communication, 5G network and LPWAN multipath communication system.

[0016] Compared with existing technologies, the beneficial effects of the multimodal collaborative medical emergency resource scheduling system proposed in this invention are as follows: it realizes the collection and fusion of multiple data sources through the resource information integration module, and combined with the emergency demand perception module, it uses big data analysis and machine learning technology to realize intelligent analysis of resource demand, and supports task tracking feedback and resource usage tracking, ensuring the accuracy and timeliness of resource scheduling; at the same time, it realizes the full-process safety visualization management of emergency response through the rescue process collaborative monitoring module. Attached Figure Description

[0017] Figure 1 This is a functional block diagram of the present invention. Detailed Implementation

[0018] The technical solution protected by this invention will be described in detail below with reference to the accompanying drawings.

[0019] This invention proposes a multimodal collaborative medical emergency resource scheduling system, comprising a resource information integration module, an emergency demand perception module, a multimodal collaborative scheduling engine, a rescue process collaborative monitoring module, a contingency plan management module, and a collaborative communication module. Through the resource information integration module and with the aid of the collaborative communication module, data from multiple data sources is collected and integrated for subsequent resource demand analysis and the generation of resource scheduling plans. The emergency demand perception module quantifies the priority of medical resource needs based on the type of emergency event, combined with a three-tiered assessment system of event severity, resource gap, and social impact. The multimodal collaborative scheduling engine, based on the priority of medical resource needs and hospital medical resources (bed availability, equipment usage, and staff scheduling, etc.), ambulance information (location, usage, etc.), and the availability of drones and volunteers, uses the NSGA-II algorithm to generate a multi-objective optimization scheme that minimizes resource transportation time and maximizes coverage of critically ill patients. The output includes collaborative scheduling instructions encompassing ambulance route planning, drone airspace routes, and volunteer task allocation.

[0020] Furthermore, the multimodal collaborative scheduling engine can dynamically optimize and correct the generated multi-objective optimization scheme based on the task execution status and the deviation between resource usage and predicted values, so as to ensure the accuracy of the model.

[0021] The rescue process collaborative monitoring module dynamically displays the execution status of the rescue plan; the contingency plan management module pre-sets three types of special contingency plans for natural disasters, accidents, and public health emergencies. Each type of contingency plan is associated with a specific resource scheduling strategy through predefined resource matching rules; the functions of each module and how they work together will be described in detail below.

[0022] When the resource information integration module receives a corresponding emergency event, it first obtains the corresponding special plan from the contingency plan management module according to the event type. Based on the special plan, it integrates the medical institution's HIS system, ambulance GPS positioning, population flow heat map and historical epidemic data through federated learning technology to construct a digital twin of medical resources, and collects dynamic status data of ambulances, drone medical units, mobile emergency units and volunteer resources in real time.

[0023] In this embodiment, the contingency plan management module pre-sets resource scheduling strategies for three types of special contingency plans: natural disasters, accidents, and public health emergencies. For example, for natural disasters such as floods, earthquakes, and typhoons, plans for disaster warning, emergency response levels, rescue team deployment, and material allocation are pre-set. For instance, for earthquakes, priority is given to allocating ICU beds and emergency equipment, and helicopters and ambulances are preferred as transportation methods, with the highest disaster warning and emergency response levels. For accidents such as fires, priority is given to allocating fire brigades and emergency vehicles, with correspondingly higher allocation levels. For public health emergencies such as epidemics, priority is given to allocating drones or logistics vehicles, as well as epidemic prevention materials and medical personnel.

[0024] The resource information integration module can acquire HIS system data of corresponding hospitals within a certain range, as well as location information of ambulances, drones, and volunteers, based on the area where the emergency occurs, in order to construct a digital twin of the corresponding medical resources. At the same time, it can utilize dynamic data collected by drones, ambulances, and volunteers, including meteorological data and data on the injuries of the injured.

[0025] The emergency demand perception module is used to quantify the priority of medical resource demand based on the type of emergency event and a three-tiered assessment system combining event severity, resource gap, and social impact. Specifically, the emergency demand perception module includes a transmission path tracking unit and an emergency room traffic prediction unit. The transmission path tracking unit is used to predict the risk of virus transmission in different regions, where the transmission risk is a risk level quantified based on the infectious disease transmission level and the number of potential infections. The emergency room traffic prediction unit is used to predict emergency room traffic based on influenza transmission areas and population flow data, where the emergency room traffic includes a resource gap level quantified based on bed and medical equipment occupancy rates and the number of critically ill patients.

[0026] More specifically, the transmission path tracking unit integrates operator roaming data, traffic checkpoint records, and public place monitoring with traffic checkpoint data and meteorological parameters based on a sequential convex programming algorithm to construct a personnel movement trajectory map. It also establishes a case contact network through a graph database, such as Nebula Graph, to identify potential transmission chains. Combined with real-time meteorological sensor data, it generates a virus transmission risk heat map based on the spatial diffusion characteristics of the Pacejka tire model.

[0027] The emergency room traffic prediction unit integrates electronic medical records, drug sales, social media sentiment, and meteorological data. Based on an improved SEIR spatiotemporal propagation model and an LSTM time series model, and superimposed with the ARIMA algorithm to capture seasonal fluctuations, it predicts the peak demand for outpatient visits within 72 hours. The prediction results are verified by the hospital's HIS system bed occupancy rate and medical equipment utilization rate. In one embodiment, the number of historical poisoning cases and weekly sales changes of antidotes are obtained from electronic medical records. By collecting frequency statistics of keywords related to social media hot topics and meteorological data such as humidity, a SEIR-LSTM-ARIMA hybrid model is used to predict the resource demand within 72 hours, obtaining the corresponding demand for beds and drugs.

[0028] The urgent need perception module combines a three-tiered assessment system of event severity, resource gap, and social impact to quantify the priority of medical resource needs, including: Based on the heat map of virus transmission risk, areas that meet the set risk level are marked as areas awaiting rescue, and priority rescue areas corresponding to the areas awaiting rescue are determined. Obtain 72-hour peak medical visit forecast information for areas awaiting rescue and corresponding priority rescue areas, and calculate the corresponding resource gap level; Based on the resource gap level of the corresponding region, the priority of medical resource demand is quantified by combining the severity of the event, the resource gap rate, and the social impact. The severity of the event is a weighted value of the quantified mortality rate, the severe illness rate, and meteorological data. The resource gap rate is the percentage of the difference between the predicted value and the actual available value of medical resources. The social impact includes corresponding indices quantified based on social media sentiment and traffic control.

[0029] Furthermore, the priority rescue areas corresponding to the areas awaiting rescue include: Obtain information on medical institutions within the designated area to be rescued; Based on the established evaluation criteria, organizations that meet the criteria are marked as priority rescue areas; The evaluation indicators include the completion rate of treatment within the golden hour and the abnormal turnover rate of supplies.

[0030] The multimodal collaborative scheduling engine connects the resource information integration module and the emergency demand perception module to acquire digital twin data, demand prediction results, and event priority instructions. It uses the NSGA-II algorithm to generate a multi-objective optimization scheme that aims to minimize resource transportation time and maximize critical care coverage, and outputs collaborative scheduling instructions that include ambulance route planning, UAV airspace routes, and volunteer task allocation.

[0031] Further, the multimodal collaborative scheduling engine includes a scheduling feedback optimization unit and a scheduling instruction execution unit; the scheduling feedback optimization unit is used to optimize and adjust the collaborative scheduling instructions according to the task execution situation and historical events; the scheduling instruction execution unit is used to decompose the collaborative scheduling instructions of the multi-objective optimization scheme generated according to the priority of medical resource requirements into executable instructions and send them to the corresponding execution terminals.

[0032] Specifically, the scheduling scheme optimization unit monitors the resource consumption through Internet of Things sensors, calculates the deviation from the predicted value, and when the actual consumption differs from the predicted value by more than the predicted value, dynamically triggers the correction of the scheduling scheme and updates the demand prediction model parameters through federated learning. For example, according to the consumption of the antidote monitored by the Internet of Things sensors, when the deviation from the predicted value exceeds the set threshold, dynamic triggering of correction is performed, including dispatching additional drones for resupply and updating the parameters of the prediction model through federated learning.

[0033] The execution process of the scheduling instruction execution unit includes: sending navigation paths and target hospital information to ambulance terminals; sending material delivery coordinates and emergency obstacle avoidance instructions to drone terminals; pushing light injury patient assembly points and safety protection guides to volunteer terminals. In order to ensure the accuracy and security of instruction transmission, instructions are distributed through a collaborative communication module. For example, the command center initiates a video conference, verifies the identity of the corresponding hospital负责人 through voiceprint verification, conveys path update instructions to ambulances through the 5G network, sends drone replenishment coordinates through satellite communication, and adjusts the volunteer assembly points through LPWAN broadcast (coverage radius 10 km).

[0034] The rescue process collaborative monitoring module integrates a GIS geographic information system and a real-time video communication unit, and is used to dynamically display the location trajectories of rescue resources, injury heat maps, and compare and mark the critical 10-minute node for on-site first aid and the critical 1-hour node for transportation; for example, using a "10-minute countdown": automatically reminding to inject the antidote on-site (5 cases of timeout warnings); "1-hour countdown": remaining time for transporting critically injured patients.

[0035] The above are only embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. All equivalent structural or equivalent process transformations made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, are equally included in the patent protection scope of the present invention.

Claims

1. A multimodal collaborative medical emergency resource scheduling system, characterized in that, include: The resource information integration module is used to integrate the HIS system of medical institutions, GPS positioning of ambulances, population flow heat maps and historical epidemic data through federated learning technology to build a digital twin of medical resources and collect dynamic status data of ambulances, drone medical units, mobile emergency rescue units and volunteer resources in real time. The emergency demand perception module is used to quantify the priority of medical resource demand based on the type of emergency event and a three-tier assessment system that combines event severity, resource gap, and social impact. The multimodal collaborative scheduling engine connects the resource information integration module and the emergency demand perception module. It uses the NSGA-II algorithm to generate a multi-objective optimization scheme that aims to minimize resource transportation time and maximize the coverage rate of severe cases. It outputs collaborative scheduling instructions that include ambulance route planning, UAV airspace routes and volunteer task allocation. The rescue process collaborative monitoring module integrates a GIS geographic information system and a real-time video communication unit to dynamically display the location trajectory of rescue resources, injury heat maps, and compare and mark the golden 10 minutes for on-site first aid and the golden 1 hour for transfer. The contingency plan management module pre-sets three types of special contingency plans: natural disasters, accidents, and public health emergencies. Each type of contingency plan is associated with a specific resource scheduling strategy through predefined resource matching rules.

2. The multimodal collaborative medical emergency resource scheduling system according to claim 1, characterized in that, The emergency demand perception module includes a transmission path tracking unit and an emergency room traffic prediction unit. The transmission path tracking unit is used to predict the risk of virus transmission in different regions. The transmission risk is a risk level quantified based on the infectious disease transmission level and the number of potential infections. The emergency room traffic prediction unit is used to predict emergency room traffic based on influenza transmission areas and population flow data. The emergency room traffic includes a resource gap level quantified based on the occupancy rate of beds and medical equipment and the number of critically ill patients.

3. The multimodal collaborative medical emergency resource scheduling system according to claim 2, characterized in that, The transmission path tracking unit integrates operator roaming data, traffic checkpoint records, and public place monitoring. Based on the sequential convex programming algorithm, it fuses traffic checkpoint data with meteorological parameters to construct a personnel movement trajectory map. Combined with real-time meteorological sensor data, it generates a virus transmission risk heat map through the spatial diffusion characteristics of the Pacejka tire model.

4. The multimodal collaborative medical emergency resource scheduling system according to claim 3, characterized in that, The emergency room traffic prediction unit integrates electronic medical records, drug sales, social media sentiment, and meteorological data. Based on the improved SEIR spatiotemporal propagation model and LSTM time series model, and superimposed with the ARIMA algorithm to capture seasonal fluctuations, it predicts the peak of outpatient visits in 72 hours. The prediction results are verified by the hospital HIS system bed occupancy rate and medical equipment utilization rate.

5. A multimodal collaborative medical emergency resource scheduling system according to claim 4, characterized in that, Based on the type of emergency, a three-tiered assessment system combining the severity of the event, resource gaps, and social impact is used to quantify the priority of medical resource needs, including: Based on the heat map of virus transmission risk, areas that meet the set risk level are marked as areas awaiting rescue, and priority rescue areas corresponding to the areas awaiting rescue are determined. Obtain 72-hour peak medical visit forecast information for areas awaiting rescue and corresponding priority rescue areas, and calculate the corresponding resource gap level; Based on the resource gap level of the corresponding region, the priority of medical resource demand is quantified by combining the severity of the event, the resource gap rate, and the social impact. The severity of the event is a weighted value of the quantified mortality rate, the critical illness rate, and meteorological data; the resource gap rate is the percentage of the difference between the predicted and actual available medical resources; and the social impact includes corresponding indices quantified based on social media sentiment and traffic control measures.

6. A multimodal collaborative medical emergency resource scheduling system according to claim 5, characterized in that, The priority rescue areas corresponding to the areas awaiting rescue include: Obtain information on medical institutions within the designated area to be rescued; Based on the established evaluation criteria, organizations that meet the criteria are marked as priority rescue areas; The evaluation indicators include the completion rate of treatment within the golden hour and the abnormal turnover rate of supplies.

7. A multimodal collaborative medical emergency resource scheduling system according to claim 1, characterized in that, The multimodal collaborative scheduling engine includes a scheduling feedback optimization unit and a scheduling instruction execution unit; the scheduling feedback optimization unit is used to optimize and adjust the collaborative scheduling instructions based on the task execution status and historical events; The scheduling instruction execution unit is used to decompose the collaborative scheduling instructions of the multi-objective optimization scheme generated according to the priority of medical resource demand into executable instructions and send them to the corresponding execution terminals.

8. A multimodal collaborative medical emergency resource scheduling system according to claim 7, characterized in that, The scheduling scheme optimization unit monitors resource consumption through IoT sensors, calculates the deviation from the predicted value, and dynamically triggers the correction of the scheduling scheme when the actual consumption differs from the predicted value by more than the predicted value, and updates the parameters of the demand prediction model through federated learning.

9. A multimodal collaborative medical emergency resource scheduling system according to claim 7, characterized in that, The execution process of the scheduling instruction execution unit includes: Send navigation routes and target hospital information to the ambulance terminal; Send the coordinates of the material delivery and emergency obstacle avoidance instructions to the drone terminal; Push information on the assembly points for lightly injured patients and safety guidelines to the volunteer terminals.

10. A multimodal collaborative medical emergency resource scheduling system according to any one of claims 1 to 9, characterized in that, It also includes a collaborative communication module, which is used to achieve one-click distribution of task instructions based on voiceprint authentication through satellite communication, 5G network and LPWAN multipath communication system.

Images