A method, system, and storage medium and device for siting an emergency device
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN SMARTCITY TECH DEV GRP CO LTD
- Filing Date
- 2025-10-30
- Publication Date
- 2026-07-03
Smart Images
Figure CN121034577B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information processing technology, and in particular to a method, system, storage medium, and device for selecting the location of emergency medical equipment. Background Technology
[0002] Automated external defibrillators (AEDs) are essential devices for rescuing patients with cardiac arrest. Deploying them in certain urban areas enables emergency treatment of out-of-hospital cardiac arrest (OHCA). This necessitates the deployment of AEDs in areas of the city where OHCA incidents are frequent.
[0003] Existing AED device site selection methods primarily rely on factors such as historical emergency data, population density, and distribution of medical resources. Specifically, based on historical OHCA (Outpatient Collapse and Acute Respiratory Syndrome) event data, Geographic Information System (GIS) tools are used to calculate event density and conduct hotspot analysis, identifying high-incidence areas of OHCA events, and maximizing AED site selection by maximizing population coverage. Summary of the Invention
[0004] This invention provides a method, system, storage medium, and device for selecting emergency medical equipment, thereby improving the rationality and accuracy of emergency medical equipment deployment.
[0005] One embodiment of the present invention provides a method for selecting the location of emergency medical equipment, comprising:
[0006] Acquire multi-dimensional data of the target area; the multi-dimensional data includes: current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data;
[0007] The target area will be gridded to obtain multiple grid cells, and grid data of each grid cell will be obtained based on the multi-dimensional data. The grid data includes current emergency medical equipment data, traffic data, weather data, social data and historical out-of-hospital emergency event data of the corresponding grid cell.
[0008] Based on the grid data, obtain the grid features of each grid cell, the spatial connection features between the multiple grid cells, and the temporal features of the multiple grid cells;
[0009] The grid features, spatial connectivity features, and temporal features are fused to obtain fused features. Based on the fused features, the occurrence information of outpatient acute illness events corresponding to each grid unit is predicted.
[0010] Based on the occurrence information of the out-of-hospital emergency events and the coverage model of the pre-set emergency equipment, the deployment information of the emergency equipment in the target area is determined.
[0011] Another embodiment of the present invention provides a location selection system for emergency medical equipment, comprising:
[0012] The data acquisition unit is used to acquire multi-dimensional data of the target area; the multi-dimensional data includes: current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data;
[0013] A gridded unit is used to perform gridding processing on the target area to obtain multiple grid units, and to obtain grid data of each grid unit based on the multi-dimensional data. The grid data includes current emergency equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data of the corresponding grid unit.
[0014] The feature acquisition unit is used to acquire the grid features of each grid cell, the spatial connection features between the multiple grid cells, and the temporal features of the multiple grid cells based on the grid data.
[0015] The prediction unit is used to fuse the grid features, spatial connectivity features, and temporal features to obtain fused features, and to predict the occurrence information of outpatient acute illness events corresponding to each grid unit based on the fused features.
[0016] The deployment unit is used to determine the deployment information of the emergency medical equipment in the target area based on the occurrence information of the out-of-hospital emergency medical event and the coverage model of the pre-set emergency medical equipment.
[0017] Another aspect of the present invention provides a computer-readable storage medium storing a plurality of computer programs adapted for loading by a processor and executing the location method for emergency medical devices as described in one aspect of the present invention.
[0018] Another embodiment of the present invention provides a terminal device, including a processor and a memory;
[0019] The memory is used to store multiple computer programs, which are loaded by a processor and executed as described in one aspect of the emergency medical device location method of the present invention; the processor is used to implement each of the multiple computer programs.
[0020] As can be seen, in the method of this embodiment, the emergency medical equipment location system performs gridding processing on the target area to obtain multiple grid cells and acquires multi-dimensional grid data for each grid cell. Based on this, it acquires the grid features, spatial connectivity features, and temporal features of each grid cell. After fusing these features, it predicts the occurrence information of out-of-hospital emergency events corresponding to each grid cell based on the fused features. Then, combined with the coverage model of emergency medical equipment, it determines the deployment information of emergency medical equipment in the target area. This process comprehensively considers multiple dimensions of factors such as time, space, weather, traffic, and society to predict out-of-hospital emergency events. Based on this, the emergency medical equipment is planned, which can be done more rationally and accurately. Attached Figure Description
[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 This is a schematic diagram of a site selection method for emergency medical equipment provided in an embodiment of the present invention;
[0023] Figure 2 This is a flowchart of a site selection method for emergency medical equipment provided in a specific embodiment of the present invention;
[0024] Figure 3a This is a schematic diagram of a risk prediction model in a specific embodiment of the present invention;
[0025] Figure 3b This is another schematic diagram of the wind direction prediction model in a specific embodiment of the present invention;
[0026] Figure 4 This is a flowchart of a site selection method for emergency medical equipment provided in an application embodiment of the present invention;
[0027] Figure 5 This is a schematic diagram of the logical structure of a location selection system for emergency medical equipment provided in an embodiment of the present invention;
[0028] Figure 6 This is a schematic diagram of the logical structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation
[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0030] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0031] This invention provides a method for selecting the location of emergency medical equipment, primarily used for site selection when deploying emergency medical equipment (such as an AED) in a target area. Figure 1 As shown, the location of emergency medical equipment, such as AEDs, is mainly determined by the emergency equipment location system using the following methods:
[0032] The process involves: acquiring multi-dimensional data of the target area; performing gridding on the target area to obtain multiple grid cells; acquiring grid data for each grid cell based on the multi-dimensional data; acquiring grid features, spatial connectivity features, and temporal features of each grid cell based on the grid data; fusing the grid features, spatial connectivity features, and temporal features to obtain fused features; predicting the occurrence information of out-of-hospital emergency events corresponding to each grid cell based on the fused features; and determining the deployment information of the emergency equipment in the target area based on the occurrence information of the out-of-hospital emergency events and a pre-set coverage model of emergency equipment.
[0033] This allows for precise analysis of the target area using grid cells as the analysis unit. By combining the spatiotemporal characteristics of the grid cells, information on out-of-hospital emergency events can be accurately predicted, thereby improving the accuracy of emergency equipment deployment in each grid cell.
[0034] A specific embodiment of the present invention provides a method for selecting the location of emergency medical equipment, mainly the method executed by the emergency medical equipment location system, the flowchart of which is shown below. Figure 2 As shown, it includes:
[0035] Step 101: Obtain multi-dimensional data of the target area.
[0036] It is understood that when initiating the process of this embodiment, the following multi-dimensional data can be obtained, but is not limited to: current emergency medical equipment data, traffic data, weather data, and social data, and may also include historical out-of-hospital emergency event data, etc., wherein:
[0037] Historical data on out-of-hospital emergency events can be obtained from emergency medical services and includes information such as the time and location of the event, medical emergency response time, and underlying medical conditions.
[0038] Current emergency medical equipment data can include the latitude and longitude, address information, attributes, and status information of existing emergency medical equipment, as well as Point of Interest (POI) data for potential emergency medical equipment locations. The emergency medical equipment mentioned here mainly refers to equipment deployed for out-of-hospital acute illnesses requiring on-site emergency assistance, such as AED devices.
[0039] Traffic data for the target area can be obtained through intelligent transportation systems, including traffic flow and road congestion.
[0040] Weather data for the target area can be obtained from meteorological departments, including historical weather conditions such as wind, rain, and snow.
[0041] Social data for the target area: administrative divisions, location point of interest (POI) data, real-time population distribution, and demographic data such as population density.
[0042] Step 102 involves performing gridding on the target area to obtain multiple grid cells, and obtaining grid data for each grid cell based on multi-dimensional data. The grid data includes current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data for the corresponding grid cell.
[0043] Specifically, a city information modeling (CIM) platform can be used to grid the target area, with each grid cell representing a small geographic region. The size of each grid cell can be specified based on the scale of the target area, data density, and actual needs. The resulting multiple grid cells need to cover all geographic points in the target area, typically 100m x 100m or 200m x 200m.
[0044] CIM, based on technologies such as Building Information Modeling (BIM), Geographic Information System (GIS), and Internet of Things (IoT), integrates multi-dimensional and multi-scale spatial data and IoT sensing data of the city's above-ground and underground, indoor and outdoor, historical, current and future, to construct an organic complex of urban information in a three-dimensional digital space.
[0045] Furthermore, when acquiring grid data for each grid cell, a unique identifier (ID) can be set for each grid cell, and attributes such as population density, traffic flow, and the number of historical outpatient emergency events can be obtained based on the aforementioned multi-dimensional data, thus realizing the gridding of the aforementioned multi-dimensional data.
[0046] It should be noted that before gridding multi-dimensional data, preprocessing can be performed on the multi-dimensional data, which can specifically include:
[0047] Data cleaning can include, but is not limited to, the following data cleaning processes: handling missing values and removing outliers;
[0048] Multi-dimensional data undergoes multi-dimensional normalization processes, such as time normalization, spatial normalization, and time alignment, to obtain normalized data.
[0049] In this way, when obtaining the grid data of each grid cell, the multidimensional standardized data corresponding to each grid cell can be obtained from the normalized data.
[0050] Specifically, when handling missing values, interpolation, averaging, or other methods can be used to fill in the missing data. For example, if traffic flow data for a certain point in time is missing, it can be filled in by averaging the traffic flow data for the preceding and following points in time.
[0051] When removing outliers, outliers (such as impossible high-traffic data) can be identified through statistical analysis and visualization, and then corrected or deleted according to reasonable rules.
[0052] During time normalization, the timestamps of all data can be unified to the same time zone and converted into a uniform time format (such as Unix timestamps).
[0053] During spatial normalization, addresses can be converted to CGCS2000 unified latitude and longitude coordinates (geocoding) for subsequent spatial analysis.
[0054] In time alignment processing, in order to integrate multi-dimensional data, it is necessary to align according to timestamps. For example, aligning traffic data and out-of-hospital emergency event data within the same time period.
[0055] Step 103: Based on the grid data obtained in Step 102 above, obtain the grid features of each grid cell, the spatial connection features between multiple grid cells, and the temporal features of multiple grid cells.
[0056] Specifically, the acquired grid features can be a spatial feature matrix, the spatial connectivity features can be a spatial adjacency matrix, and the temporal features refer to the feature values of each grid cell at different times. Wherein:
[0057] When acquiring grid features, the grid data obtained above can be used to obtain static and dynamic features. Static features can include grid cell traffic flow, population density, age, distance from medical institutions, actual geographical location, historical out-of-hospital emergency event data such as emergency response time, distribution of people with heart disease, distribution of people with hypertension, distribution of people with diabetes, distribution of existing emergency equipment, opening hours of areas where emergency equipment is located, and average income level of people. Dynamic features can include grid cell real-time traffic flow and current weather conditions.
[0058] Specifically, the static and dynamic characteristics of any grid cell can be normalized, and then the normalized static and dynamic characteristics can be weighted and averaged to obtain the grid characteristics of the grid cell.
[0059] When acquiring spatial connectivity features, the connection relationships between various grid cells and the weights of these connections can be established to identify the strength of mutual influence between grid cells. Specifically, this weighting can be calculated based on the geographical distance between grid cells, with closer cells having greater weights. Furthermore, this weighting can be dynamically adjusted based on data such as traffic flow and population movement.
[0060] The edge index matrix edge_index can be obtained based on the connection relationship between grid cells, and the edge weight matrix edge_weights can be obtained based on the weight of the connection between grid cells.
[0061] When acquiring temporal features, different time periods can be constructed. Specifically, a week can be divided into two groups: weekdays and rest days. Each group can be further divided into multiple time periods at one-hour intervals. Then, the grid features and spatial connectivity features of the grid cells in these multiple time periods can be collected to obtain the temporal features.
[0062] Specifically, an array representing the deployment of emergency medical equipment can be obtained based on temporal characteristics. Each element in the array indicates whether an emergency medical device is available in a grid cell within a given time period. If it is, the element's value is 1; otherwise, the value is 0. It's important to note that the availability of emergency medical equipment here primarily considers two factors: the service status of the equipment (normal or abnormal) and the service status of the location where the equipment is located, taking into account commuting hours and business hours to define its service time range. Only when the emergency medical equipment's service status is normal and the location's service status is open will the corresponding element in the array have a value of 1.
[0063] Step 104: The grid features, spatial connectivity features, and temporal features are fused to obtain fused features. Based on the fused features, the occurrence information of outpatient emergency events corresponding to each grid unit is predicted.
[0064] Specifically, a pre-set risk prediction model for out-of-hospital emergency events can be invoked. The risk prediction model integrates grid features, spatial connectivity features, and temporal features to obtain fused features. Based on the fused features, the probability of occurrence of out-of-hospital emergency events corresponding to each grid unit can be predicted, i.e., the probability of occurrence of out-of-hospital emergency events.
[0065] The risk prediction model is a machine learning model based on artificial intelligence (AI). AI refers to the theories, methods, technologies, and application systems that use digital computers or computers-controlled machines to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess perception, reasoning, and decision-making capabilities.
[0066] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, machine learning, and deep learning.
[0067] Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instruction-based learning.
[0068] In specific instances, such as Figure 3a As shown, the risk prediction model may include a spatial feature layer 10, a spatiotemporal sequence layer 11, and an output layer 12, wherein:
[0069] Spatial feature layer 10 is used to perform convolution processing on the mesh features and spatial connectivity features of any mesh cell to obtain spatial convolution features;
[0070] The spatiotemporal sequence layer 11 is used to fuse the temporal features and spatial convolutional features of the grid cells to obtain the fused features;
[0071] Output layer 12 is used to output the probability of out-of-hospital emergency events in grid cells based on the fused features. For example, when the probability of occurrence exceeds 0.4, the grid cell should be equipped with emergency medical equipment, and when the probability of occurrence exceeds 0.7, the grid cell is considered a high-risk area and multiple emergency medical equipment need to be deployed.
[0072] Furthermore, such as Figure 3b As shown, the risk prediction model may also include an attention layer 13, which is specifically used to process the spatial convolution features using an attention mechanism to obtain spatial attention features. It mainly focuses on certain important features in the spatial convolution features, enhances these important features, and weakens other unimportant features. In this way, the aforementioned spatiotemporal sequence layer 11 will fuse the spatial attention features and temporal features to obtain fused features.
[0073] In the specific implementation process, the spatial feature layer 10 can be a spatiotemporal graph neural network (ST-GNN), which is specifically used to process spatiotemporal data, such as traffic flow prediction, weather forecasting, social network analysis, etc. It combines the characteristics of graph neural networks (GNN) and recurrent neural networks (RNN) and can simultaneously capture the spatial dependence and temporal dynamics in the data.
[0074] The spatiotemporal sequence layer 11 can be a bidirectional long short-term memory network (LSTM) or a gated recurrent unit (GRU) to process the temporal series correlation of spatiotemporal features and enhance the ability to model time series data.
[0075] It should be noted that each grid cell can correspond to a risk prediction model, allowing multiple risk prediction models to be processed in parallel, each predicting the probability of outpatient acute events in its respective grid cell. This improves the efficiency of predicting outpatient acute events across multiple grid cells.
[0076] Step 105: Based on the occurrence information of out-of-hospital emergency events and the coverage model of pre-set emergency equipment, determine the deployment information of each emergency equipment in the target area.
[0077] It should be noted that the coverage model for emergency medical equipment is primarily based on the principle of minimizing emergency response time or maximizing high-risk areas to plan the location of emergency medical equipment, thereby achieving reasonable deployment of AED devices and efficient emergency response. This coverage model can utilize multi-objective optimization algorithms, such as Multi-Objective Particle Swarm Optimization (MOPSO) or Genetic Algorithm (GA), to obtain deployment information of emergency medical equipment in some grid cells within the target area. Specifically:
[0078] Based on information about out-of-hospital emergency events, potential grid units for the deployment of emergency medical equipment are set, such as areas with high traffic such as shopping malls, schools, bus stations, and supermarkets, as well as grid units where the probability of out-of-hospital emergency events exceeds a set value.
[0079] Based on the coverage model that maximizes the risk area criterion, the service radius of emergency medical equipment in potential grid cells is set to R; or, based on the coverage model that minimizes emergency medical response time criterion, the deployment information of emergency medical equipment in potential grid cells is set.
[0080] This involves iteratively calculating service radii from different location selection algorithms, selecting an algorithm whose calculated service radius covers areas with a high probability of out-of-hospital emergency events, and then using the selected algorithm to obtain the final service radius. These location selection algorithms can include multi-objective optimization algorithms, such as Multi-Objective Particle Swarm Optimization (MOPSO) or Genetic Algorithm (GA), to ensure that high-risk areas (areas with a high probability of out-of-hospital emergency events) can quickly access emergency medical services.
[0081] Alternatively, based on a coverage model that minimizes emergency response time, the shortest time cost to reach the deployment point of emergency equipment in the grid cell can be calculated under different location algorithms. The location algorithm that calculates the shortest time cost for AED devices can be selected, and the deployment information of emergency equipment can be obtained based on the selected location algorithm to improve the success rate of emergency rescue.
[0082] Specifically, let's take the multi-objective particle swarm optimization algorithm as an example to illustrate this:
[0083] Initialize the particle swarm, which consists of multiple particles, each corresponding to a grid cell, and randomly assign an initial position and velocity to each particle;
[0084] The fitness value of each particle is calculated, and the Pareto front method is used to compare and store the non-dominated solutions of the optimal solution group.
[0085] Update the position and velocity of each particle:
[0086] The optimal solution for each particle and the global optimal solution are updated based on the fitness function. The final solution is selected based on actual needs, and the deployment information of emergency rescue equipment in each grid cell is generated based on the optimization algorithm results.
[0087] Furthermore, the deployment information of identified emergency medical equipment in grid cells can be visualized through the CIM platform.
[0088] As can be seen, in the method of this embodiment, the emergency medical equipment location system performs gridding processing on the target area to obtain multiple grid cells and acquires the grid data of each grid cell. Based on this, it acquires the grid features, spatial connection features, and temporal features of each grid cell. After fusing these features, it predicts the occurrence information of out-of-hospital emergency events corresponding to each grid cell based on the fused features. Then, combined with the coverage model of emergency medical equipment, it determines the deployment information of emergency medical equipment in the target area. This process comprehensively considers multiple dimensions of factors such as time, space, weather, traffic, and society to predict out-of-hospital emergency events. Based on this, the emergency medical equipment is planned, which can be done more rationally and accurately.
[0089] It should be noted that, in order to predict the occurrence information of outpatient acute illness events corresponding to each grid cell in the above embodiments, a risk prediction model can be trained in advance. Specifically, when training the risk prediction model:
[0090] Construct a basic risk prediction model, which may include the methods described above. Figure 3a The spatial feature layer, spatiotemporal sequence layer, and output layer shown above, or as described above... Figure 3b The structure shown;
[0091] Acquire training samples, which include: grid features, spatial connectivity features, and temporal features of multiple sample grids in the sample region, as well as information on outpatient acute illness events that have occurred in the history of each sample grid;
[0092] The original risk prediction model is trained based on the training samples to obtain the above-mentioned pre-set risk prediction model.
[0093] The method for obtaining the grid features, spatial connectivity features, and temporal features of multiple sample grids in the sample region included in the training samples is similar to the method for obtaining the grid features, spatial connectivity features, and temporal features of each grid cell in the target region, and will not be elaborated here.
[0094] Specifically, when training the original risk prediction model, it can first fuse the grid features, spatial connectivity features, and temporal features of each sample grid, and then predict the probability of an out-of-hospital emergency event occurring in each sample grid based on the fused features. A loss function, such as the binary cross-entropy loss function, is then set to measure the error between the probability predicted by the original risk prediction model and the historical information on out-of-hospital emergency events occurring in each sample grid. A gradient is then calculated based on the value of the loss function, and the Adam optimization algorithm is used to optimize the parameters, adjusting the learning rate and regularization parameters to prevent overfitting.
[0095] Furthermore, the K-fold cross-validation method can be used to adjust the hyperparameters in the original risk prediction model, such as the learning rate, kernel size, number of graph convolutional layers, and number of temporal convolutional layers.
[0096] By repeatedly adjusting the parameters in the original risk prediction model, the loss function is made to meet certain conditions. At this point, a risk prediction model is obtained and can be pre-installed in the system so that step 104 above can be executed when the location selection process for emergency medical equipment is initiated.
[0097] The following specific application example illustrates the location method for emergency medical equipment in this implementation. In this embodiment, the emergency medical equipment is an AED device, and the out-of-hospital emergency event is an OHCA event, as an example. Figure 4 As shown, it includes:
[0098] Step 201: Obtain multi-dimensional data of the sample area, which may include historical OHCA event data, AED data, traffic data, weather data, and social data.
[0099] Step 202 involves performing gridding on the sample area to obtain multiple sample grids, and obtaining the grid data of each sample grid based on multi-dimensional data.
[0100] Step 203: Obtain training samples. The training samples include: obtaining the grid features of each sample grid based on the grid data obtained in step 202 above, obtaining the spatial connectivity features between multiple sample grids, obtaining the temporal features of multiple sample grids, and obtaining information on historical OHCA events for each sample grid.
[0101] Step 204: Construct the original risk prediction model. The structure of the original risk prediction model can be as described above. Figure 3a or Figure 3b As shown.
[0102] Step 205: The grid features of each sample grid, the spatial connectivity features between multiple sample grids, and the time series data are fused through the original risk prediction model to obtain fused features. Based on the fused features, the probability of occurrence of OHCA events corresponding to each sample grid is predicted. Then, a loss function is set according to the probability of occurrence of OHCA events corresponding to the sample grids and the information of historical OHCA events in the corresponding sample grids in the training samples. The hyperparameters in the original risk prediction model are optimized based on the loss function to obtain the trained risk prediction model.
[0103] After training the risk prediction model, the risk prediction model is pre-installed into the AED device location system. When the location process for AED devices in the target area is initiated, steps 206 to 209 can be executed.
[0104] Step 206: Obtain multi-dimensional data of the target area, which may include historical OHCA event data, AED data, traffic data, weather data, and social data.
[0105] Step 207 involves performing gridding on the target area to obtain multiple grid cells, acquiring grid data for each grid cell based on multi-dimensional data, and then acquiring grid features of each grid cell, spatial connection features between multiple grid cells, and temporal features of multiple grid cells based on the grid data.
[0106] Step 208: Invoke the preset OHCA event risk prediction model, and fuse grid features, spatial connectivity features and temporal features through the risk prediction model to obtain fused features. Based on the fused features, predict the occurrence probability of OHCA events corresponding to each grid cell.
[0107] Step 209: Based on the occurrence probability of OHCA events and the pre-set coverage model of AED devices, determine the deployment information of each AED device in the target area.
[0108] As can be seen, the AED device location method in this embodiment can achieve the following technical effects:
[0109] In risk prediction models, temporal and spatial features are simultaneously introduced into graph neural networks to form a spatiotemporal graph structure, enabling the model to capture the complex interaction between OHCA events in time and space, and thus better predict dynamically changing risks.
[0110] High-precision OHCA event risk prediction: By using a risk prediction model to analyze data in both time and space dimensions, high-risk OHCA areas can be predicted more accurately, generating detailed risk heat maps that allow planners to intuitively understand the OHCA risk distribution in each region.
[0111] In the process of selecting AED device locations, multiple optimization objectives are comprehensively considered, such as response time, coverage, and traffic flow. Based on actual application scenarios and environmental changes, a model is proposed to minimize emergency response time and maximize coverage in high-risk areas, thereby determining the optimal facility layout scheme and ensuring the practicality and flexibility of the location selection plan.
[0112] This invention also provides a location selection system for emergency medical equipment, the structural diagram of which is shown below. Figure 5 As shown, it can specifically include:
[0113] The data acquisition unit 20 is used to acquire multi-dimensional data of the target area. The multi-dimensional data includes: current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data.
[0114] The gridding unit 21 is used to perform gridding processing on the target area to obtain multiple grid units, and to obtain grid data of each grid unit based on the multi-dimensional data obtained by the data acquisition unit 20. The grid data includes current emergency equipment data, traffic data, weather data, social data and historical out-of-hospital emergency event data of the corresponding grid unit.
[0115] When acquiring grid data from each grid cell, the grid unit 21 specifically performs time normalization, spatial normalization, and time alignment processing on the multi-dimensional data to obtain normalized data; sets a unique identifier for each grid cell, and obtains multi-dimensional standardized data corresponding to each grid cell based on the normalized data.
[0116] The feature acquisition unit 22 is used to acquire the grid features of each grid cell, the spatial connection features between the multiple grid cells, and the temporal features of the multiple grid cells based on the grid data acquired by the gridding unit 21.
[0117] The prediction unit 23 is used to fuse the grid features, spatial connectivity features and temporal features acquired by the feature acquisition unit 22 to obtain fused features, and predict the occurrence information of outpatient emergency events corresponding to each grid unit based on the fused features.
[0118] The prediction unit 23 is specifically used to call a preset risk prediction model for out-of-hospital acute illness events; the risk prediction model is used to fuse the grid features, spatial connectivity features and temporal features to obtain fused features, and the probability of occurrence of out-of-hospital acute illness events corresponding to each grid unit is predicted based on the fused features.
[0119] The risk prediction model may include a spatial feature layer, a spatiotemporal sequence layer, and an output layer, wherein: the spatial feature layer is used to perform convolution processing on the grid features and spatial connectivity features of any grid cell to obtain spatial convolution features; the spatiotemporal sequence layer is used to fuse the temporal features and spatial convolution features to obtain fused features; and the output layer is used to output the probability of occurrence of outpatient acute illness events for the corresponding grid cell based on the fused features.
[0120] Furthermore, the risk prediction model further includes: an attention layer for processing the spatial convolutional features using an attention mechanism to obtain spatial attention features; and a spatiotemporal sequence layer for fusing the spatial attention features and temporal features to obtain fused features.
[0121] Deployment unit 24 is used to determine the deployment information of the emergency medical equipment in the target area based on the occurrence information of out-of-hospital emergency events predicted by prediction unit 23 and the pre-set coverage model of emergency medical equipment.
[0122] The deployment unit 24 is specifically used to set potential grid units for the deployment of emergency medical equipment based on the occurrence information of the out-of-hospital emergency illness events; and to set the service radius of the emergency medical equipment in the potential grid units based on a coverage model that maximizes the risk area criterion, or to set the deployment information of the emergency medical equipment in the potential grid units based on a coverage model that minimizes the emergency response time criterion.
[0123] Furthermore, the location system for the emergency medical equipment in this embodiment may also include:
[0124] Training unit 25 is used to construct a risk prediction original model, which includes a spatial feature layer, a spatiotemporal sequence layer and an output layer; to acquire training samples, which include: grid features, spatial connectivity features and temporal sequence features of multiple sample grids in the sample area, and information on outpatient acute illness events that have occurred in the history of each sample grid; and to train the risk prediction original model based on the training samples to obtain the preset risk prediction model.
[0125] In the system of this embodiment, this process comprehensively considers multiple dimensions of factors such as time, space, weather, traffic, and society to predict out-of-hospital emergency events. Based on this, emergency equipment is planned, which can make the planning of emergency equipment more reasonable and accurate.
[0126] This invention also provides a terminal device, the structural schematic of which is shown below. Figure 6 As shown, the terminal device can vary significantly due to differences in configuration or performance. It may include one or more central processing units (CPUs) 30 (e.g., one or more processors) and memory 31, and one or more storage media 32 (e.g., one or more mass storage devices) for storing application programs 321 or data 322. The memory 31 and storage media 32 can be temporary or persistent storage. The program stored in the storage media 32 may include one or more modules (not shown in the figure), each module including a series of instruction operations on the terminal device. Furthermore, the CPU 30 may be configured to communicate with the storage media 32 and execute the series of instruction operations in the storage media 32 on the terminal device.
[0127] Specifically, the application program 321 stored in the storage medium 32 includes an application program for the location selection of emergency medical devices. This program may include the data acquisition unit 20, the gridding unit 21, the feature acquisition unit 22, the prediction unit 23, the deployment unit 24, and the training unit 25 in the aforementioned emergency medical device location selection system, which will not be elaborated here. Furthermore, the central processing unit 30 may be configured to communicate with the storage medium 32 and execute a series of operations corresponding to the emergency medical device location selection application program stored in the storage medium 32 on the terminal device.
[0128] The terminal device may also include one or more power supplies 33, one or more wired or wireless network interfaces 34, one or more input / output interfaces 35, and / or one or more operating systems 323, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.
[0129] The steps performed by the emergency medical equipment location system in the above method embodiments can be based on this. Figure 6 The structure of the terminal device shown is illustrated.
[0130] Furthermore, embodiments of the present invention also provide a computer-readable storage medium storing a plurality of computer programs adapted for loading by a processor and executing a method for selecting emergency medical devices as described above.
[0131] Furthermore, embodiments of the present invention also provide a terminal device, including a processor and a memory;
[0132] The memory is used to store multiple computer programs, which are loaded by a processor and executed as described above, representing a method for selecting emergency medical devices; the processor is used to implement each of the multiple computer programs.
[0133] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.
[0134] The above provides a detailed description of the site selection method, system, storage medium, and device for an emergency medical device provided by the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method of siting an emergency device, characterized by, The emergency medical equipment mentioned is an out-of-hospital emergency medical equipment, including: Acquire multi-dimensional data of the target area; the multi-dimensional data includes: current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data; The target area will be gridded to obtain multiple grid cells, and grid data of each grid cell will be obtained based on the multi-dimensional data. The grid data includes current emergency medical equipment data, traffic data, weather data, social data and historical out-of-hospital emergency event data of the corresponding grid cell. Based on the grid data, obtain the grid features of each grid cell, the spatial connection features between the multiple grid cells, and the temporal features of the multiple grid cells; Specifically, obtaining the spatial connection features between the multiple grid cells includes: establishing the connection relationship between each grid cell and generating an edge index matrix; and obtaining the weight of the connection between each grid cell based on the geographical distance between each grid cell and generating an edge weight matrix. The acquisition of the temporal features of each grid cell specifically includes: constructing multiple time periods, collecting grid features and spatial connectivity features of grid cells under the multiple time periods; obtaining an array to represent the deployment of emergency medical equipment based on the temporal features, wherein each element in the array represents whether there is available emergency medical equipment in a grid cell within a time period; The grid features, spatial connectivity features, and temporal features are fused to obtain fused features. The occurrence information of out-of-hospital acute illness events corresponding to each grid unit is predicted based on the fused features. Specifically, when fusing the grid features, spatial connectivity features, and temporal features to obtain fused features, the grid features and spatial connectivity features of any grid unit are convolved through a spatiotemporal graph neural network to obtain spatial convolution features. Then, the temporal features and spatial convolution features of the grid unit are fused through a spatiotemporal sequence layer to obtain fused features. Based on the occurrence information of the out-of-hospital emergency events and the coverage model of the pre-set emergency equipment, the deployment information of the emergency equipment in the target area is determined; The step of determining the deployment information of the emergency medical equipment in the target area based on the occurrence information of the out-of-hospital emergency medical events and the pre-set coverage model of the emergency medical equipment specifically includes: Based on the occurrence information of the out-of-hospital emergency illness events, potential grid units for the deployment of the emergency medical equipment are set; Based on a coverage model that maximizes the risk area criterion, the service radius of emergency medical equipment in the potential grid cell is set; or based on a coverage model that minimizes emergency response time criterion, the deployment information of emergency medical equipment in the potential grid cell is set. The service radius is obtained by iterating different location algorithms. The location algorithm whose calculated service radius can cover areas with a high probability of out-of-hospital emergency events is selected, and the service radius is obtained based on the selected location algorithm. The coverage model based on the criterion of minimizing emergency response time calculates the shortest time cost to reach the deployment point of emergency equipment in the network unit under different selection algorithms. The location algorithm of emergency equipment with the shortest time cost is selected, and the deployment information of the emergency equipment is obtained based on the selected location algorithm.
2. The method of claim 1, wherein, The step of obtaining the grid data of each grid cell based on the multi-dimensional data specifically includes: The multi-dimensional data is subjected to time normalization, spatial normalization, and time alignment to obtain normalized data; Each grid cell is assigned a unique identifier, and multidimensional standardized data for each grid cell is obtained based on the normalized data.
3. The method of claim 1, wherein, The process of fusing the grid features, spatial connectivity features, and temporal features to obtain fused features, and predicting the occurrence information of outpatient acute illness events corresponding to each grid unit based on the fused features, specifically includes: Invoke the pre-set risk prediction model for outpatient emergency events; The risk prediction model fuses the grid features, spatial connectivity features, and temporal features to obtain fused features, and predicts the probability of occurrence of out-of-hospital acute illness events corresponding to each grid unit based on the fused features.
4. The method of claim 3, wherein, The risk prediction model includes: a spatial feature layer, a spatiotemporal sequence layer, and an output layer, wherein: The spatial feature layer is used to perform convolution processing on the grid features and spatial connectivity features of any grid cell to obtain spatial convolution features; The spatiotemporal sequence layer is used to fuse the temporal features and spatial convolutional features to obtain fused features; The output layer is used to output the probability of occurrence of outpatient acute illness events for the corresponding grid cells based on the fused features.
5. The method of claim 4, wherein, The risk prediction model also includes: The attention layer is used to process the spatial convolutional features using an attention mechanism to obtain spatial attention features; The spatiotemporal sequence layer is used to fuse the spatial attention features and temporal features to obtain fused features.
6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: A risk prediction primitive model is constructed, which includes a spatial feature layer, a spatiotemporal sequence layer, and an output layer. Acquire training samples, which include: grid features, spatial connectivity features and temporal features of multiple sample grids in the sample region, and information on outpatient acute illness events that have occurred in the history of each sample grid; The original risk prediction model is trained based on the training samples to obtain a preset risk prediction model.
7. A siting system for an emergency device, characterized in that The emergency medical equipment mentioned is an out-of-hospital emergency medical equipment, including: The data acquisition unit is used to acquire multi-dimensional data of the target area; the multi-dimensional data includes: current emergency medical equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data; A gridded unit is used to perform gridding processing on the target area to obtain multiple grid units, and to obtain grid data of each grid unit based on the multi-dimensional data. The grid data includes current emergency equipment data, traffic data, weather data, social data, and historical out-of-hospital emergency event data of the corresponding grid unit. A feature acquisition unit is used to acquire the grid features of each grid cell, the spatial connection features between the multiple grid cells, and the temporal features of the multiple grid cells based on the grid data. Specifically, acquiring the spatial connection features between the multiple grid cells includes: establishing connection relationships between the grid cells and generating an edge index matrix; and obtaining the weights of the connections between the grid cells based on the geographical distance between them, generating an edge weight matrix. Acquiring the temporal features of each grid cell specifically includes: constructing multiple time periods and collecting the grid features and spatial connection features of the grid cells in the multiple time periods; and obtaining an array representing the deployment of emergency medical equipment based on the temporal features, where each element in the array represents whether there is available emergency medical equipment in a grid cell within a time period. The prediction unit is used to fuse the grid features, spatial connectivity features, and temporal features to obtain fused features, and to predict the occurrence information of outpatient acute illness events corresponding to each grid unit based on the fused features. The prediction unit is specifically used to perform convolution processing on the grid features and spatial connectivity features of any grid cell through a spatiotemporal graph neural network to obtain spatial convolution features, and then perform fusion processing on the temporal features and spatial convolution features of the grid cell through a spatiotemporal sequence layer to obtain fused features. The deployment unit is used to determine the deployment information of the emergency medical equipment in the target area based on the occurrence information of the out-of-hospital emergency medical event and the coverage model of the pre-set emergency medical equipment; The deployment unit is specifically used to set potential grid units for the deployment of emergency medical equipment based on the occurrence information of the out-of-hospital emergency illness events; to set the service radius of the emergency medical equipment in the potential grid units based on a coverage model that maximizes the risk area criterion, or to set the deployment information of the emergency medical equipment in the potential grid units based on a coverage model that minimizes the emergency response time criterion; wherein, the service radius is obtained by iterating different location algorithms, selecting a location algorithm whose calculated service radius can cover areas with a high probability of out-of-hospital emergency illness events, and obtaining the service radius based on the selected location algorithm; the coverage model based on the minimum emergency response time criterion calculates the shortest time cost to reach the deployment point of the emergency medical equipment in the network unit under different selection algorithms, selects the location algorithm for the emergency medical equipment that calculates the shortest time cost, and obtains the deployment information of the emergency medical equipment based on the selected location algorithm.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of computer programs adapted to be loaded by a processor and executed by the location method for emergency medical devices as described in any one of claims 1 to 6.
9. A terminal device, characterized in that, Including processor and memory; The memory is used to store a plurality of computer programs, which are loaded by a processor and executed as described in any one of claims 1 to 6, the processor being used to implement each of the plurality of computer programs.