Sanitary emergency personal protection grading method and personal protection article issuing system

Abstract

The invention discloses a health emergency personal protection grading method and a personal protection article issuing system, and belongs to the technical field of personal protection. The personal protection grading method comprises the following steps: acquiring task information data and epidemic situation information data; the epidemic situation information data comprises current epidemic intensity data, infectious data and hazard data of infectious diseases; the task information data comprises epidemic area exposure duration, task area average population density and task activity range radius; and performing calculation based on the epidemic situation information data and the task information data by using the trained grading model to obtain the personal protection grade. According to the method provided by the invention, the real-time condition of the epidemic situation is fully considered, various indexes of the epidemic situation are organically combined with the actual needs of the emergency task, and the personal protection level is calculated, so that the emergency personnel are guaranteed to obtain enough personal protection, waste caused by excessive protection is avoided, and the utilization rate of materials is improved.

Description

Technical Field

[0001] The present invention belongs to the field of personal protection technology, and in particular relates to a health emergency personal protection grading method and a personal protective equipment distribution system. Background Art

[0002] In the event of a sudden health emergency caused by an infectious disease outbreak, relevant departments must be able to respond quickly and handle the situation in a scientific and effective manner. Personal protection is the first line of defense for emergency personnel and a prerequisite for ensuring the smooth implementation of subsequent work. Currently, the grading and assessment criteria for personal protection standards are relatively simple. There is no mechanism for real-time adjustment based on the changing epidemic situation, nor is there targeted optimization based on the actual needs of emergency response tasks. This leads to confusion in the selection and distribution of protective equipment in different scenarios, often resulting in wasted resources or inadequate protection levels. Summary of the Invention

[0003] In view of this, the present invention provides a health emergency personal protection classification method and a personal protective equipment distribution system to improve the reliability of personal protection classification and increase the utilization rate of materials in emergency events.

[0004] In order to achieve the above object, the solution adopted by the present invention is: a grading method for personal protection in health emergencies, comprising the following steps:

[0005] Obtain task information data and epidemic information data; the epidemic information data includes the current epidemic intensity data, infectiousness data and harmfulness data of the infectious disease; the task information data includes the exposure time in the epidemic area, the average population density in the task area and the radius of the task activity range; use the trained grading model to calculate the personal protection level based on the epidemic information data and task information data.

[0006] Furthermore, the epidemic intensity data includes at least one of the following dimensions: incidence rate, prevalence rate and attack rate.

[0007] Furthermore, the infectiousness data includes at least one of the following dimensions: infection rate, secondary attack rate, basic reproduction number, effective reproduction number and case growth rate.

[0008] Furthermore, the hazard data includes at least one of the following dimensions: case fatality rate, severity rate and mortality rate.

[0009] Furthermore, the task information data is first preprocessed and then input into the hierarchical model; the preprocessing includes: performing cube root transformation processing on one or more of the data including the exposure time data of the epidemic area, the average population density data of the task area and the radius data of the task activity range.

[0010] Furthermore, the internal calculation process of the hierarchical model includes:

[0011] The epidemic intensity data and the radius of the mission activity range are input into the first integration module to calculate the first feature set; the infectious data and the average population density of the mission area are input into the second integration module to calculate the second feature set; the hazard data and the exposure time in the epidemic area are input into the third integration module to calculate the third feature set; the first feature set, the second feature set and the third feature set are spliced ​​row by row to obtain the first comprehensive feature set; the first comprehensive feature set is calculated in sequence by the first convolution layer, the first activation layer, the tail fully connected layer and the softmax classifier to obtain the personal protection level.

[0012] Furthermore, the first integration module, the second integration module and the third integration module are all fully connected neural networks.

[0013] The hierarchical model of the present invention first uses multiple integration modules to disperse and integrate task information data and epidemic information data, and then calculates the results through the first convolutional layer, the first activation layer, the tail fully connected layer and the softmax classifier. Compared with the conventional single fully connected network, the model has less calculation amount, is easier to train, and has a better fitting effect on the multi-dimensional data in the task information data and epidemic information data.

[0014] Furthermore, the epidemic information data also includes historical time series data of infectious diseases, and the historical time series data includes the following dimensions: infection rate, incidence rate, basic reproduction number and effective reproduction number.

[0015] Furthermore, a fourth integration module is provided within the grading model. The fourth integration module calculates a fourth feature set based on the historical time series data. The fourth feature set is fused with the data output by the first activation layer, and then calculated through the tail fully connected layer and the softmax classifier to obtain the personal protection level.

[0016] The internal calculation process of the fourth integration module includes:

[0017] Convolution calculations are performed on the historical time series data of each dimension to obtain multiple historical feature sets; the multiple historical feature sets are spliced ​​row by row to obtain a second comprehensive feature set; the second comprehensive feature set is calculated in sequence through the second convolutional layer, the second activation layer, the middle fully connected layer and the third activation layer to obtain the fourth feature set.

[0018] The present invention also provides a personal protective equipment distribution system, comprising a memory and a processor; the memory is configured to store processor-executable instructions, and the processor is connected to the memory; the processor is configured to execute the method described above to obtain a personal protection level. The memory may also store a list of supplies corresponding to each protection level. The processor may also be electrically connected to a processor in an automatic protective equipment distribution terminal. After calculating the personal protection level, the system obtains the list of supplies corresponding to the corresponding level based on the stored information. To ensure that the size of the protective equipment distributed matches the size of the emergency personnel, the memory of the protective equipment distribution system also stores the individual body size data of emergency personnel with different identities. Before distributing protective equipment, the emergency personnel enter their identity information (via facial recognition, keyboard input, card swiping, etc.) into the distribution system. The system then determines the model and size of the various protective equipment items in the list of supplies based on the stored body size data. Finally, the list of supplies containing the protective equipment models is transmitted to the automatic protective equipment distribution terminal, and the corresponding personal protective equipment is automatically distributed according to the list of supplies.

[0019] This solution allows health emergency personnel to quickly prepare personal protective equipment (PPE), eliminating the need for manual record keeping and preventing errors in searching and selecting PPE when collecting it. It also facilitates emergency supply management, ensuring a first-in-first-out system and eliminating the need for regular inventory checks. This reduces the labor costs associated with distributing and checking emergency supplies. Compared to traditional methods of collecting PPE, this solution shortens the time required, accelerating emergency response deployments.

[0020] The beneficial effects of the present invention are:

[0021] In sudden emergency situations, there are often varying degrees of shortages of manpower and material resources. Making full use of resources and making efficient and accurate decisions will be of great help in controlling the epidemic. The method provided by the present invention fully considers the real-time situation of the epidemic and organically combines various indicators of the epidemic with the actual needs of emergency tasks, thereby calculating the personal protection level. This not only ensures that emergency personnel obtain sufficient personal protection, but also avoids waste caused by excessive protection, and improves the utilization rate of materials, which is especially important when personnel and materials are in short supply.

[0022] After the present invention uses data labeled by human experts to train the hierarchical model, the model can learn the relevant knowledge contained in this data information. Inexperienced personnel can quickly obtain professional personal protection reference information through this method, which helps improve their ability to respond to the epidemic. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 The figure is a flow chart of the personal protection classification method for health emergencies of the present invention. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of the technical solutions of this application, but not all of them. Based on the embodiments described in this application document, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the technical solutions of this application.

[0025] Example 1: Figure 1 As shown, a grading method for personal protection in health emergencies includes the following steps:

[0026] Obtain mission information data and current epidemic information data. Specifically, health emergency mission types include epidemiological investigations, disinfection and pest control, vector monitoring, sample collection, regional lockdowns, and material deployment.

[0027] Epidemic information data includes the current epidemic intensity data, infectiousness data, and hazard data of infectious diseases. Epidemic intensity data includes at least one of the following dimensions: incidence rate, prevalence rate, and attack rate; infectiousness data includes at least one of the following dimensions: infection rate, secondary attack rate (SAR), basic reproduction number (Ro), effective reproduction number (Rt), and growth rate of case numbers; hazard data includes at least one of the following dimensions: case fatality rate (CFR), severity rate, and mortality rate.

[0028] The mission information data includes the duration of exposure to the epidemic area (in hours), the average population density of the mission area (in people / km²), and the radius of the mission activity range (in kilometers). The duration of exposure to the epidemic area and the radius of the mission activity range can be clearly obtained based on the received mission information. Specifically, the radius of the mission activity range refers to the distance from the center point of the activity range at the time of mission execution to the farthest edge of the activity range. In some embodiments, the average population density of the mission area can be determined based on the mission activity range. Depending on the size of the activity range, the average population density of the mission area is usually determined based on existing statistical data for administrative regions such as villages, towns, districts, counties, cities, and provinces. For example, if the mission range is within a certain district in a certain city, the average population density of the mission area is the average population density of that district. For another example, if the mission range is within a certain township, the average population density of the mission area is the average population density of that township. In other embodiments, the average population density of the mission area can also be calculated independently. For example, if the mission range is limited to a certain school, hospital, or factory, the data can be calculated based on the area and population of the school, hospital, or factory.

[0029] The aforementioned task information and epidemic information data are input into the trained grading model, which then calculates the individual protection level based on the epidemic information and task information data. When training the model using a training set, the loss function can be a cross-entropy loss function, and the labeled data in the training dataset comes from historical empirical data and / or labeled data annotated by human experts.

[0030] As one of the preferred implementation methods, the task information data is first preprocessed and then input into the trained classification model. Specifically, the preprocessing includes: performing a cube root transformation on one or more of the data including the exposure time data of the epidemic area, the average population density data of the task area, and the radius data of the task activity range, so as to reduce the impact of the dimensional differences of the data in different dimensions and improve the accuracy and stability of the model classification. Specifically, the cube root transformation processing is performed according to the following formula:

[0031]

[0032] In the above formula, N represents the task information data before preprocessing, and M represents the data after cube root transformation.

[0033] The classification model can be implemented using an existing nonlinear data fitting model. As one preferred embodiment, the classification model includes a first integration module, a second integration module, a third integration module, and a fourth integration module. The calculation process within the classification model includes:

[0034] The epidemic intensity data and the radius of the mission activity range are input into the first integration module to calculate the first feature set; the infectiousness data and the average population density of the mission area are input into the second integration module to calculate the second feature set; the hazard data and the exposure time in the epidemic area are input into the third integration module to calculate the third feature set; the first feature set, the second feature set and the third feature set are spliced ​​row by row to obtain the first comprehensive feature set; the first comprehensive feature set is calculated in sequence by the first convolution layer, the first activation layer, the tail fully connected layer and the softmax classifier to obtain the personal protection level.

[0035] The first integration module, the second integration module, and the third integration module are all fully connected neural networks. Sigmoid is used as the activation function within each of the fully connected neural networks. In some embodiments, each integration module includes a sequentially connected input layer, two hidden layers, and an output layer. The number of input nodes in the input layer of each integration module is determined according to the input data dimension. The output layer of each integration module outputs vectors of the same length, namely, the first feature set, the second feature set, and the third feature set. The first comprehensive feature set is a two-dimensional matrix with a height of 3 and a width equal to the length of each feature set. The convolution kernel size of the first convolution layer can be 3*3. During the convolution operation, the convolution window slides along the width direction of the first comprehensive feature set. The output result after the first convolution layer operation is a one-dimensional vector. The first activation layer preferably uses the PReLU function.

[0036] Example 2: For some infectious diseases, they have been prevalent many times in different regions and at different times. At present, a lot of historical time series data related to these diseases have been accumulated. These historical time series data reflect the changes in the relevant dimensions of the infectious disease over time. When a certain infectious disease breaks out again, its development often has many similarities with the previous ones. Therefore, the accumulated historical time series data can be used to characterize and predict the future development direction of the current epidemic. Specifically, the historical time series data can include four dimensions: historical time series data of infection rate, historical time series data of incidence rate, historical time series data of basic reproduction number, and historical time series data of effective reproduction number. These historical time series data cover information from the early stage of each epidemic to the end of the epidemic. As one of the implementation methods, when constructing the training set, the original historical time series data is sampled at equal intervals according to a preset frequency. The sampling frequency is determined according to the length of each time series data. The more time series data there is, the lower the sampling frequency is. The length of all historical time series data finally obtained is equal.

[0037] In this embodiment, the hierarchical model also includes a fourth integration module. This module calculates a fourth feature set based on historical time series data. This feature set is then fused with the data output by the first activation layer. The data is then passed through the tail fully connected layer and a softmax classifier to calculate the personal protection level. This allows the model to learn from historical time series data how the evolution of infectious diseases over time affects personal protection levels. The output is based not only on the current epidemic situation but also on future changes in the epidemic. This makes the hierarchical model's output more adaptable and reliable for upcoming tasks.

[0038] As one preferred embodiment, the calculation process within the fourth integration module includes:

[0039] Convolution calculations are performed on the historical time series data of each dimension to obtain four historical feature sets (convolution calculations are performed on the historical time series data of one dimension to obtain one historical feature set). Since historical time series data is a vector, the convolution kernel size can be 1*1*K when convolving the historical time series data, where K is equal to the length of the historical time series data, and the lengths of the multiple historical feature sets obtained are equal. Multiple historical feature sets are concatenated row by row to obtain a second comprehensive feature set, which is a two-dimensional matrix with a height of 4.

[0040] The second comprehensive feature set is calculated in sequence through the second convolutional layer, the second activation layer, the middle fully connected layer and the third activation layer to obtain the fourth feature set, which is a one-dimensional vector data. Among them, the convolution kernel size of the second convolutional layer can be 4*4, and the convolution window slides along the width direction of the second comprehensive feature set. The result output after the second convolutional layer operation is a vector. The second activation layer preferably uses the PReLU function, and the third activation layer preferably uses the sigmoid function. In some embodiments, the length of the fourth feature set is equal to the length of the vector output by the first activation layer, and the fourth feature set is fused by multiplying the corresponding elements of the data output by the first activation layer.

[0041] In one preferred implementation, both the epidemic information data and historical time series data in the training set come from infectious diseases with similar transmission modes (e.g., both are airborne). When the tiered model is deployed to perform personal protection tiers, the infectious diseases targeted have similar transmission modes to those in the training set (e.g., also airborne), which further improves the accuracy of the results.

[0042] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A grading method for personal protection in health emergencies, characterized by: The following steps are involved: Obtain task information data and epidemic information data; The epidemic information data includes the current epidemic intensity data, infectiousness data and harmfulness data of the infectious disease; The mission information data includes the duration of exposure to the epidemic area, the average population density of the mission area, and the radius of the mission activity range; The trained grading model is used to calculate the personal protection level based on the epidemic information data and the task information data.

2. The method for grading personal protection in health emergencies according to claim 1, characterized in that: The epidemic intensity data includes at least one of the following dimensions: incidence rate, prevalence rate and attack rate.

3. The method for grading personal protection in health emergencies according to claim 1, characterized in that: The infectiousness data includes at least one of the following dimensions: infection rate, secondary attack rate, basic reproduction number, effective reproduction number and case growth rate.

4. The method for grading personal protection in health emergencies according to claim 1, characterized in that: The hazard data includes at least one of the following dimensions: case fatality rate, severity rate and mortality rate.

5. The method for grading personal protection in health emergencies according to claim 1, characterized in that: The task information data is first preprocessed and then input into the hierarchical model; the preprocessing includes: performing cube root transformation on one or more of the data of exposure duration in the epidemic area, average population density in the task area, and radius of the task activity range.

6. The method for grading personal protection in health emergencies according to any one of claims 1 to 5, characterized in that: The internal calculation process of the hierarchical model includes: Inputting the epidemic intensity data and the task activity range radius into a first integration module to calculate and obtain a first feature set; Inputting the infectious data and the average population density of the mission area into a second integration module to calculate a second feature set; Inputting the hazard data and the duration of exposure to the epidemic area into a third integration module to calculate a third feature set; Concatenate the first feature set, the second feature set, and the third feature set row by row to obtain a first comprehensive feature set; The first comprehensive feature set is calculated in sequence through the first convolutional layer, the first activation layer, the tail fully connected layer and the softmax classifier to obtain the personal protection level.

7. The method for grading personal protection in health emergencies according to claim 6, characterized in that: The first integration module, the second integration module and the third integration module are all fully connected neural networks.

8. The method for grading personal protection in health emergencies according to claim 6, characterized in that: The epidemic information data also includes historical time series data of infectious diseases, and the historical time series data includes the following dimensions: infection rate, incidence rate, basic reproduction number and effective reproduction number.

9. The method for grading personal protection in health emergencies according to claim 8, characterized in that: The grading model is further provided with a fourth integration module, which calculates a fourth feature set based on the historical time series data. The fourth feature set is fused with the data output by the first activation layer, and then calculated through the tail fully connected layer and the softmax classifier to obtain the personal protection level. The internal calculation process of the fourth integration module includes: Perform convolution calculations on the historical time series data of each dimension to obtain multiple historical feature sets; Multiple historical feature sets are concatenated row by row to obtain a second comprehensive feature set; The second comprehensive feature set is calculated in sequence by the second convolutional layer, the second activation layer, the middle fully connected layer and the third activation layer to obtain the fourth feature set.

10. A personal protective equipment distribution system, characterized in that: It comprises a memory and a processor; the memory is used to store processor-executable instructions, and the processor is connected to the memory; the processor is configured to execute the method according to any one of claims 1 to 9 to obtain a personal protection level, and then distribute corresponding personal protective equipment according to the personal protection level.

Images