Health emergency monitoring method and system under heavy pollution weather

Abstract

The present application discloses a method and system for health emergency monitoring under heavy pollution weather, which involves health emergency, data processing, and artificial intelligence, including: establishing a map mark of a target area on a corresponding regional map according to the location of each health monitoring point; determining possible monitoring blind spots; configuring a drone flight path according to possible monitoring blind spots; obtaining health monitoring data covering the target area from health monitoring points and configured drones when the indicators of weather monitoring data exceed the prediction threshold; preprocessing the health monitoring data, and using an improved ConvLSTM model to predict the possible distribution of health events in the future based on the preprocessed health monitoring data; performing hierarchical marking based on the regional map according to the prediction results, and executing corresponding health emergency strategies for the hierarchically marked areas. The method of the present application can effectively improve the auxiliary management effect of health emergency.

Description

Technical Field

[0001] The present application relates to the fields of health emergency, data processing, and artificial intelligence technology, and in particular to a health emergency monitoring method and system under heavy pollution weather. Background Art

[0002] For densely populated areas, it is very important to understand the health conditions in the region, especially when the environment is deteriorating. By monitoring regional health conditions, the response speed of health centers and disease control centers can be improved, and possible regional health events can be controlled in advance.

[0003] Traditional health area monitoring methods rely heavily on fixed monitoring sites and lack the ability to dynamically schedule effective monitoring equipment, resulting in the inability to cover monitoring blind spots in a timely manner. In particular, in scenarios where pollution spreads rapidly, data collection is significantly delayed. Summary of the invention

[0004] The embodiments of the present application provide a health emergency monitoring method and system under heavy pollution weather, which improves the prediction accuracy of the propagation path of pollutants through an improved ConvLSTM model, thereby effectively improving the auxiliary effect of health emergency.

[0005] The present application embodiment proposes a health emergency monitoring method under heavy pollution weather, including:

[0006] Pre-set multiple fixed health monitoring points, and establish map marks of the target area on the corresponding regional map according to the location of each health monitoring point;

[0007] Determine the area that each health monitoring point can cover, and determine possible monitoring blind spots based on the regional map;

[0008] Configure the UAV flight path based on possible monitoring blind spots;

[0009] Acquire weather monitoring data of a target area, and when an indicator of the weather monitoring data exceeds a prediction threshold, acquire health monitoring data covering the target area from health monitoring points and configured drones;

[0010] Preprocessing the health monitoring data, and using the improved ConvLSTM model to predict the possible distribution of future health events based on the preprocessed health monitoring data;

[0011] According to the prediction results, the regional map is graded and marked, and corresponding health emergency strategies are implemented for the graded and marked areas.

[0012] Optionally, determining the area that each health monitoring point can cover and determining possible monitoring blind areas based on the regional map includes:

[0013] Determine the geographical elevation of each health monitoring point and determine the average elevation of the target area;

[0014] According to the geographical elevation of any health monitoring point and the average elevation, a diffusion coefficient is configured for the health monitoring point, wherein the diffusion coefficient of the health monitoring point at the average elevation is set to 1, the diffusion coefficient of the monitoring point with a geographical elevation higher than the average elevation is greater than 1, and the diffusion coefficient of the monitoring point with a geographical elevation lower than the average elevation is less than 1;

[0015] Determine the coverage area of ​​each health monitoring point according to the reference diffusion area and the configured diffusion coefficient, and determine the uncovered area based on the regional map;

[0016] Uncovered areas are screened to obtain possible monitoring blind spots.

[0017] Optionally, based on possible monitoring blind spots, configure the drone flight path including:

[0018] Obtaining the geographic elevation of the possible monitoring blind area;

[0019] According to the geographical elevation of the monitoring blind area, dynamic monitoring points are configured in the corresponding monitoring blind area, and the monitoring data of the dynamic monitoring points covers the monitoring blind area, otherwise the number of dynamic monitoring points is increased;

[0020] Configure the initial path of the drone based on all dynamic monitoring points in each monitoring blind area;

[0021] The distribution positions of the dynamic monitoring points are adjusted to optimize the initial path to obtain a flight path.

[0022] Optionally, obtaining health monitoring data covering the target area from the health monitoring point and the configured drone includes:

[0023] Determine the pollutant information monitored from the health monitoring points;

[0024] Determining associated pollutants based on the determined pollutant information, and acquiring associated pollutant information covering the target area through a drone; and,

[0025] Obtain meteorological data, traffic data, and medical data in the target area as auxiliary monitoring data.

[0026] Optionally, preprocessing the health monitoring data includes:

[0027] The obtained relevant data are interpolated by Kriging and converted into grid data based on the target area to align the multi-source data. The Kriging interpolation process satisfies:

[0028]

[0029] in, represents the weight coefficient, is the grid point to be interpolated, is the neighboring health monitoring point or drone path coverage point of the grid point to be interpolated, and N is the total number of neighboring health monitoring points or drone path coverage points around the grid point to be interpolated.

[0030] Optionally, the improved ConvLSTM model is used to predict the distribution of possible future health events based on the preprocessed health monitoring data, including:

[0031] Configure the data time window length, the spatial dimension of the relevant data, and set the number of channels according to the relevant pollutants and auxiliary monitoring data;

[0032] Insert a graph attention network between the hidden layers of the ConvLSTM model to improve the ConvLSTM model:

[0033]

[0034] in, represents the hidden state of the l+1th layer at time t, represents the hidden state of layer l at time t, represents multi-channel data input at time t, It is a graph attention network used to process the spatial dependencies of health monitoring data. It is an LSTM combined with convolution operation to process spatiotemporal data.

[0035] Optionally, using the improved ConvLSTM model to predict the distribution of possible future health events based on the preprocessed health monitoring data also includes: fusing the auxiliary monitoring data using the dynamic weights of multimodal cross-attention.

[0036] Optionally, performing hierarchical marking based on the regional map according to the prediction result includes:

[0037] Determining gridded data having changes according to the prediction results;

[0038] Based on the current grid data and the changed grid data, a regional map with a heat map overlay is generated, and the risk areas are graded and marked according to the preset grading indicators.

[0039] An embodiment of the present application also proposes a health emergency monitoring system under heavy pollution weather, including a processor and a memory, wherein a computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the health emergency monitoring method under heavy pollution weather as described above are implemented.

[0040] The embodiment of the present application achieves full coverage of the target area by combining fixed health monitoring points and controlling the flight path of the drone, and improves the prediction accuracy of the propagation path of pollutants through the improved ConvLSTM model, thereby effectively improving the auxiliary effect of health emergency.

[0041] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0043] Figure 1 The following is a schematic diagram of the basic process of the health emergency monitoring method under heavy pollution weather of this embodiment. DETAILED DESCRIPTION

[0044] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0045] The present application embodiment proposes a health emergency monitoring method under heavy pollution weather, such as Figure 1 As shown, the following steps are included:

[0046] In step S101, a plurality of fixed health monitoring points are pre-set, and map marks of the target area are established on the corresponding regional map according to the positions of the health monitoring points, that is, the fixed health monitoring points are mapped to the regional map of the target area, and map marks can also be established for the health monitoring points.

[0047] In step S102, the area covered by each health monitoring point is determined, and possible monitoring blind areas are determined based on the regional map. In a specific example, the range that cannot be covered can be determined based on the area covered by each monitoring point, thereby determining possible monitoring blind areas.

[0048] In step S103, the flight path of the UAV is configured according to possible monitoring blind spots.

[0049] In step S104, weather monitoring data of the target area is obtained, and when the indicators of the weather monitoring data exceed the prediction threshold, health monitoring data covering the target area is obtained from health monitoring points and configured drones. That is, emergency health emergency monitoring can be initiated in severe weather, and possible monitoring blind spots can be dynamically monitored by using drones carrying probes or sensors to increase attention to health events in the target area and improve the response efficiency of regional health emergencies.

[0050] In step S105, the health monitoring data is preprocessed, and the improved ConvLSTM model is used to predict the possible future health event distribution based on the preprocessed health monitoring data. In some specific embodiments, the improved ConvLSTM model is used for prediction, and some spatiotemporal data can be introduced by improving the ConvLSTM model to improve the adaptability of the model to the target area, thereby improving the prediction accuracy.

[0051] In step S106, the regional map is graded and marked according to the prediction results, and the corresponding health emergency strategy is implemented for the graded and marked areas. For example, emergency strategy grades such as warning, measure 1, and measure 2 are pre-set, and the prediction results are superimposed on the current regional map monitoring results, so that the regional map can be marked intuitively to provide a basis for auxiliary decision-making for the regional CDC or health clinic.

[0052] In some embodiments, determining the area that can be covered by each health monitoring point and determining possible monitoring blind areas based on the regional map includes:

[0053] Determine the geographic elevation of each health monitoring point and determine the average elevation of the target area. For example, the geographic elevation can be based on an altitude of 0, and further determine the overall average elevation.

[0054] According to the geographical elevation of any health monitoring point and the average elevation, a diffusion coefficient is configured for the any health monitoring point, wherein the diffusion coefficient of the health monitoring point at the average elevation is set to 1, the diffusion coefficient of the monitoring point whose geographical elevation is higher than the average elevation is greater than 1, and the diffusion coefficient of the monitoring point whose geographical elevation is lower than the average elevation is less than 1.

[0055] In the specific example, assuming the average elevation is 50 meters, the elevation of a monitoring point A is 70 meters, and its diffusion coefficient is set to 1.2; the elevation of monitoring point B is 40 meters, and the diffusion coefficient is set to 0.8. The diffusion coefficient calculation formula is:

[0056]

[0057] in, is the diffusion coefficient, is the monitoring point elevation, is the average elevation;

[0058] The coverage area of ​​each health monitoring point is determined based on the reference diffusion area and the configured diffusion coefficient, and the uncovered area is determined based on the regional map.

[0059] The specific coverage area can be calculated by multiplying the base diffusion area (such as 5km²) by the diffusion coefficient to determine the actual coverage of each monitoring point. For example, the coverage area of ​​monitoring point A is 5×1.2=6km², and that of monitoring point B is 5×0.8=4km².

[0060] The uncovered areas are screened to obtain possible monitoring blind areas, for example, the uncovered areas (such as a suburb that is not covered due to terrain obstruction) are identified and marked as monitoring blind areas.

[0061] In some embodiments, configuring the flight path of the drone according to possible monitoring blind spots includes:

[0062] Obtaining the geographic elevation of the possible monitoring blind area;

[0063] According to the geographical elevation of the monitoring blind area, dynamic monitoring points are configured in the corresponding monitoring blind area, and the monitoring data of the dynamic monitoring points covers the monitoring blind area, otherwise the number of dynamic monitoring points is increased. For example, in some examples, multiple groups of dynamic monitoring points can be configured so that as long as the drone flies over the dynamic monitoring points of the monitoring blind area, the sampling of the monitoring blind area can be covered.

[0064] Configure the initial path of the drone based on all dynamic monitoring points in each monitoring blind area.

[0065] The distribution positions of the dynamic monitoring points are adjusted to optimize the initial path to obtain a flight path.

[0066] In some embodiments, obtaining health monitoring data covering the target area from the health monitoring point and the configured drone includes:

[0067] Determine the pollutant information monitored from the health monitoring points.

[0068] The associated pollutants are determined based on the determined pollutant information, and the associated pollutant information covering the target area is acquired through the drone. In some embodiments, for example, the pollutant information pm2.5 can be associated with pm10, etc. as the associated pollutant information.

[0069] Obtain meteorological data, traffic data, and medical data of the target area as auxiliary monitoring data. For example, the meteorological data contains data such as wind force level and rainfall that may have an impact on pollutants, and traffic data also has similar impacts. Medical data can reflect the type and trend of the current condition of patients in the target area. In the embodiment of this application, the relevant data is used as auxiliary monitoring data.

[0070] In some embodiments, preprocessing the health monitoring data includes:

[0071] The obtained relevant data is interpolated by Kriging and converted into grid data based on the target area. For example, the target area can be converted into a 500m*500m grid based on the map and elevation to align the multi-source data. The Kriging interpolation process satisfies:

[0072]

[0073] in, represents the weight coefficient, is the grid point to be interpolated, is the neighboring health monitoring point or drone path coverage point of the grid point to be interpolated, and N is the total number of neighboring health monitoring points or drone path coverage points around the grid point to be interpolated.

[0074] In some embodiments, using the improved ConvLSTM model to predict the possible future health event distribution based on the preprocessed health monitoring data includes:

[0075] Configure the data time window length, for example, the time window inputs 6 hours of historical data (72 5-minute slices), the spatial dimension of the relevant data (500m*500m grid), and set the number of channels according to the relevant pollutants and auxiliary monitoring data. For example, one channel can be set for each type of auxiliary monitoring data.

[0076] Insert a graph attention network between the hidden layers of the ConvLSTM model to improve the ConvLSTM model:

[0077]

[0078] in, represents the hidden state of the l+1th layer at time t, Indicates l The hidden state of the layer at time t, represents multi-channel data input at time t, It is a graph attention network used to process the spatial dependencies of health monitoring data. It is an LSTM combined with convolution operation to process spatiotemporal data.

[0079] In some embodiments, using the improved ConvLSTM model to predict the possible future health event distribution based on the preprocessed health monitoring data also includes: fusing the auxiliary monitoring data using the dynamic weights of multimodal cross attention to meet the following requirements:

[0080]

[0081]

[0082] in, is the weight coefficient of environmental characteristics (including meteorological data and traffic data), is the learnable weight matrix, For meteorological characteristics, For traffic characteristics, is the medical data feature, Represents a splicing operation, To fusion features, is an activation function, such as Sigmoid. Through dynamic adjustment of real-time data, the problem that traditional fixed weights cannot adapt to sudden meteorological changes is solved.

[0083] In some embodiments, performing hierarchical marking based on the regional map according to the prediction result includes:

[0084] Determining gridded data having changes according to the prediction results;

[0085] Based on the current grid data and the changing grid data, a regional map with a heat map overlay is generated, and the risk areas are graded and marked according to the preset grading indicators. For example, high-risk areas (PM2.5>250μg / m³) and corresponding emergency health measures can be marked to improve the prevention and control capabilities of health incidents.

[0086] The embodiment of the present application achieves full coverage of the target area by combining fixed health monitoring points and controlling the flight path of the drone, and improves the prediction accuracy of the propagation path of pollutants through the improved ConvLSTM model, thereby effectively improving the auxiliary effect of health emergency.

[0087] An embodiment of the present application also proposes a health emergency monitoring system under heavy pollution weather, including a processor and a memory, wherein a computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the health emergency monitoring method under heavy pollution weather as described above are implemented.

[0088] In addition, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present disclosure with equivalent elements, modifications, omissions, combinations (e.g., various embodiments intersecting schemes), adaptations or changes. It is not limited to the examples described in this specification or during the implementation of this application, and its examples will be interpreted as non-exclusive.

[0089] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. For example, those of ordinary skill in the art may use other embodiments when reading the above description.

[0090] The above embodiments are merely exemplary embodiments of the present disclosure. Those skilled in the art may make various modifications or equivalent substitutions to the present invention within the essence and protection scope of the present disclosure, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present invention.

Claims

1. A health emergency monitoring method under heavy pollution weather, characterized in that: include: Pre-set multiple fixed health monitoring points, and establish map marks of the target area on the corresponding regional map according to the location of each health monitoring point; Determine the area that each health monitoring point can cover, and determine possible monitoring blind spots based on the regional map; Configure the UAV flight path based on possible monitoring blind spots; Acquire weather monitoring data of a target area, and when an indicator of the weather monitoring data exceeds a prediction threshold, acquire health monitoring data covering the target area from health monitoring points and configured drones; Preprocessing the health monitoring data, and using the improved ConvLSTM model to predict the possible distribution of future health events based on the preprocessed health monitoring data; Marking the regions based on the prediction results and implementing corresponding health emergency strategies for the marked regions; Determine the area that each health monitoring point can cover, and determine the possible monitoring blind areas based on the regional map, including: Determine the geographical elevation of each health monitoring point and determine the average elevation of the target area; According to the geographical elevation of any health monitoring point and the average elevation, a diffusion coefficient is configured for the health monitoring point, wherein the diffusion coefficient of the health monitoring point at the average elevation is set to 1, the diffusion coefficient of the monitoring point with a geographical elevation higher than the average elevation is greater than 1, and the diffusion coefficient of the monitoring point with a geographical elevation lower than the average elevation is less than 1; Determine the coverage area of ​​each health monitoring point according to the reference diffusion area and the configured diffusion coefficient, and determine the uncovered area based on the regional map; Uncovered areas are screened for possible monitoring blind spots.

2. The health emergency monitoring method under heavy pollution weather as claimed in claim 1, characterized in that: According to the possible monitoring blind spots, the configuration of the drone flight path includes: Obtaining the geographic elevation of the possible monitoring blind area; According to the geographical elevation of the monitoring blind area, dynamic monitoring points are configured in the corresponding monitoring blind area, and the monitoring data of the dynamic monitoring points covers the monitoring blind area, otherwise the number of dynamic monitoring points is increased; Configure the initial path of the drone based on all dynamic monitoring points in each monitoring blind area; The distribution positions of the dynamic monitoring points are adjusted to optimize the initial path to obtain a flight path.

3. The health emergency monitoring method under heavy pollution weather as claimed in claim 1, characterized in that: Acquiring health monitoring data covering the target area from health monitoring points and based on the configuration of drones includes: Determine the pollutant information monitored from the health monitoring points; Determining associated pollutants based on the determined pollutant information, and acquiring associated pollutant information covering the target area through a drone; and, Obtain meteorological data, traffic data, and medical data in the target area as auxiliary monitoring data.

4. The method for health emergency monitoring under heavy pollution weather according to claim 1, characterized in that: Preprocessing the health monitoring data includes: The obtained relevant data are interpolated by Kriging and converted into grid data based on the target area to align the multi-source data. The Kriging interpolation process satisfies: ; in, represents the weight coefficient, is the grid point to be interpolated, is the neighboring health monitoring point or drone path coverage point of the grid point to be interpolated, and N is the total number of neighboring health monitoring points or drone path coverage points around the grid point to be interpolated.

5. The method for health emergency monitoring under heavy pollution weather as claimed in claim 4, characterized in that: The improved ConvLSTM model is used to predict the distribution of possible health events in the future based on the preprocessed health monitoring data, including: Configure the data time window length, the spatial dimension of the relevant data, and set the number of channels according to the relevant pollutants and auxiliary monitoring data; Insert a graph attention network between the hidden layers of the ConvLSTM model to improve the ConvLSTM model: ; in, represents the hidden state of the l+1th layer at time t, represents the hidden state of layer l at time t, represents multi-channel data input at time t, It is a graph attention network used to process the spatial dependencies of health monitoring data. It is an LSTM combined with convolution operation to process spatiotemporal data.

6. The method for health emergency monitoring under heavy pollution weather as claimed in claim 5, characterized in that: Using the improved ConvLSTM model to predict the distribution of possible future health events based on the preprocessed health monitoring data also includes: fusing the auxiliary monitoring data using the dynamic weights of multimodal cross-attention.

7. The method for health emergency monitoring under heavy pollution weather as claimed in claim 5, characterized in that: Grading marking based on the regional map according to the prediction results includes: Determining gridded data having changes according to the prediction results; Based on the current grid data and the changed grid data, a regional map with a heat map overlay is generated, and the risk areas are graded and marked according to the preset grading indicators.

8. A health emergency monitoring system under heavy pollution weather, characterized in that: It includes a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the health emergency monitoring method under heavy pollution weather as described in any one of claims 1 to 7 are implemented.

Images