Pre-hospital emergency demand prediction method and system based on multi-modal fusion

Abstract

The invention provides a pre-hospital emergency demand prediction method and system based on multi-modal fusion, and belongs to the technical field of big data multi-modal information processing. Performing feature extraction on the historical time series data of the pre-hospital emergency demand to obtain time series coding features; obtaining external factor embedding features according to external factor data corresponding to the pre-hospital emergency demand, and connecting the external factor embedding features with the time sequence coding features to obtain sequence feature representation; according to spatial image data corresponding to the pre-hospital emergency demand, spatial feature representation is obtained, the sequence feature representation and the spatial feature representation are subjected to cross-modal attention fusion, and a pre-hospital emergency demand prediction result is obtained; according to the invention, deep fusion is carried out on features from different modes through an effective data fusion means, the data dimension is fully expanded, the limitation and one-sidedness caused by single data are avoided, and the precision of a pre-hospital emergency prediction result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of big data multimodal information processing, and in particular to a method and system for predicting pre-hospital emergency demand based on multimodal fusion. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] Emergency incidents occur every day in cities. Once an emergency incident occurs, the dispatch center needs to immediately dispatch emergency resources to provide emergency services based on patient information. If a large number of emergency incidents occur in a certain area in a short period of time, it will put tremendous pressure on the emergency resources and rescue efficiency of the area. Economically developed and densely populated areas have more abundant emergency resources. Predicting the emergency needs in the future can establish a reasonable resource management method, greatly shorten the arrival time of emergency treatment, and improve the survival rate of patients. For economically backward and sparsely populated areas, emergency resources are relatively insufficient. It is even more urgent to achieve active emergency treatment by fully mining historical data to establish an emergency demand prediction model. Reasonable deployment of limited emergency resources based on the predicted regional distribution will help further improve the emergency and critical care emergency system.

[0004] The inventors found that historical emergency data usually only include brief information registration, and the data dimensions available in the pre-hospital environment are relatively few, which limits the predictive ability of the model in emergency research and may lead to biased or inaccurate prediction results; in addition, the impact of external factors on pre-hospital emergency prediction results is often ignored, resulting in low accuracy of the final prediction results. Summary of the invention

[0005] In order to address the shortcomings of the prior art, the present invention provides a method and system for predicting pre-hospital emergency demand based on multimodal fusion, which deeply integrates features from different modalities through effective data fusion means, fully expands the data dimension, avoids the limitations and one-sidedness caused by single data, and improves the accuracy of pre-hospital emergency prediction results.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for predicting pre-hospital emergency demand based on multimodal fusion.

[0007] A method for predicting pre-hospital emergency demand based on multimodal fusion includes the following processes: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; According to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series coding features to obtain the sequence feature representation; According to the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

[0008] As a further limitation of the first aspect of the present invention, feature extraction is performed on the historical time series data of pre-hospital emergency demand to obtain time series coding features, including: ; in, represents the activation function, Indicates the encoder's layer, Indicates time, For community areas, represents sparse self-attention, represents a one-dimensional convolution, Indicates time Community Area The encoder The input of the layer, represents maximum pooling, Represents time series encoding features.

[0009] As a further limitation of the first aspect of the present invention, the external factor data at least includes: weather conditions, air quality, temperature and population size corresponding to the pre-hospital emergency needs in each time period.

[0010] As a further limitation of the first aspect of the present invention, according to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding feature is obtained, including: The external factor data are concatenated and input into the stacked convolutional layers to obtain the external factor embedding features.

[0011] As a further limitation of the first aspect of the present invention, according to the spatial image data corresponding to the pre-hospital emergency demand, a spatial feature representation is obtained. ,include: ; ; ; ; ; in, represents one-dimensional convolution, represents a two-dimensional convolution, is the input of the attention module, represents channel attention, represents spatial attention, is the multiplication of pixel values, Act represents the activation function, is a normalization operation, represents the average pooling operation, represents the maximum pooling operation, Indicates time, Indicates the number of channels of the feature map, and represents the height and width of the feature map, Represents the input spatial image.

[0012] As a further limitation of the first aspect of the present invention, the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain a pre-hospital emergency demand prediction result, including: inputting the fusion result into a fully connected layer to obtain a final pre-hospital emergency demand prediction result.

[0013] In a second aspect, the present invention provides a pre-hospital emergency demand prediction system based on multimodal fusion.

[0014] A pre-hospital emergency demand prediction system based on multimodal fusion, comprising: The time series coding feature extraction module is configured to: extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; The sequence feature representation generation module is configured to: obtain external factor embedding features according to the external factor data corresponding to the pre-hospital emergency demand, and connect the external factor embedding features with the time series encoding features to obtain the sequence feature representation; The demand prediction result generation module is configured to obtain spatial feature representation based on spatial image data corresponding to the pre-hospital emergency demand, and fuse the sequence feature representation with the spatial feature representation through cross-modal attention to obtain the pre-hospital emergency demand prediction result.

[0015] In a third aspect, the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for predicting pre-hospital emergency demand based on multimodal fusion as described in the first aspect of the present invention.

[0016] In a fourth aspect, the present invention provides an electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for predicting pre-hospital emergency demand based on multimodal fusion as described in the first aspect of the present invention are implemented.

[0017] In a fifth aspect, the present invention provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the method for predicting pre-hospital emergency demand based on multimodal fusion as described in the first aspect of the present invention.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention innovatively provides a method for predicting the demand for pre-hospital emergency care based on multimodal fusion. Through effective data fusion means, the features from different modalities are deeply integrated, the data dimension is fully expanded, the limitations and one-sidedness caused by single data are avoided, and the accuracy of pre-hospital emergency care prediction results is improved.

[0019] 2. The present invention uses the transformer encoding layer to effectively extract features from historical time series data, which can effectively reduce the time complexity of the model and improve computational efficiency; the external factor data is incorporated into the model, and the extracted external factors are connected and input into the embedding layer to obtain the external factor embedding features; in order to extract spatial dependency information, a spatial feature extraction module is used to capture the global spatiotemporal features, and finally, multi-source heterogeneous data are fused at the feature level. Taking full account of the characteristics of data of different modalities, a cross-modal attention fusion module is designed to achieve reliable prediction under multi-dimensional constraints.

[0020] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0022] Figure 1 A schematic diagram of a flow chart of a method for predicting pre-hospital emergency demand based on multimodal fusion provided in Example 1 of the present invention; Figure 2 A schematic diagram of a pre-hospital emergency demand prediction system based on multimodal fusion provided in Example 2 of the present invention; Figure 3 A schematic diagram of a computer device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0024] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0025] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0026] Embodiment 1: This implementation provides a method for predicting pre-hospital emergency demand based on multimodal fusion. Through the transformer encoding layer based on the sparse self-attention mechanism, effective feature extraction is performed on historical time series data, which can effectively reduce the time complexity of the model and improve computational efficiency. In addition, the present invention incorporates external factors into the model, connects the extracted external factors and inputs them into the embedding layer to obtain external factor embedding features. In order to extract spatial dependency information, a spatial feature extraction module is used to capture global spatiotemporal features. Finally, multi-source heterogeneous data are fused at the feature level, and the characteristics of different modal data are fully considered. A cross-modal attention fusion module is designed to achieve reliable prediction under multi-dimensional constraints. The present invention can provide effective reference information for the allocation of medical resources, predict the development of regional emergency demand, and help further improve the emergency and critical care system.

[0027] Specifically, Figure 1 As shown, the following process is included: S1: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features.

[0028] More specifically, for feature extraction of time series data (including historical data on the number of emergency calls, historical population characteristics, temperature, environmental quality, etc.), the transformer time series encoding layer based on the sparse self-attention mechanism is used. It can help learn useful features in the historical stream and obtain high-level representations with dominant features.

[0029] The transformer time series encoding layer based on the sparse self-attention mechanism can be expressed as follows: (1-1); (1-2); In the formula, represents the activation function, which is Activation function, Indicates the encoder's layer, Indicates time, For community areas, represents sparse self-attention, represents a sparse matrix of the same size as the query vector, The top-u queries are restricted to keep only the query vectors with strong responses; represents a one-dimensional convolution, Indicates time Community Area The encoder The input of the layer, represents maximum pooling, is the activation function, is the key vector, Represents the extraction result of time series encoding features (if there are multiple layers, it also represents the input of the next layer i+1), represents the dimension of the key, where the query vector and key vector All passed Sure.

[0030] S2: Based on the external factor data corresponding to the demand for pre-hospital emergency care, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series encoding features to obtain the sequence feature representation.

[0031] Changes in pre-hospital emergency care are affected by many factors. For example, frequent traffic accidents in rainy and snowy environments will increase the number of trauma emergency care. Therefore, external factors such as temperature, environmental quality, rainy and snowy weather, and population characteristics are also taken into consideration.

[0032] Specifically, the present invention extracts multi-dimensional continuous external features including weather conditions, air quality, temperature and population characteristics, and splices the extracted external features into the embedding layer to obtain the external factor embedding features. , express time, represents the external characteristic length, Represents the embedding feature dimension. The external factor embedding feature is calculated by stacking convolution layers (i.e., one-dimensional convolution operation). The obtained external factor embedding feature is connected with the time series encoding feature (for example, direct splicing or weighted splicing can be used), and finally the sequence feature representation is obtained. .

[0033] S3: Based on the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

[0034] For spatial image information, the present invention implements spatial and channel attention calculation to deal with the clustering characteristics of emergency distribution in space, and this module is lightweight. Then the spatial image features are obtained through the convolution block. , specifically, including: (1-3); (1-4); (1-5); (1-6); (1-7); in, represents one-dimensional convolution, represents a two-dimensional convolution, is the input of the attention module, represents channel attention, represents spatial attention, is the pixel value multiplied, represents the activation function, is a normalization operation, represents the average pooling operation, represents the maximum pooling operation, Indicates time, Indicates the number of channels of the feature map, and represents the height and width of the feature map, Represents the input spatial image.

[0035] The feature extraction methods based on sequence and image information are completely different, and the features obtained are also generated in different domains. In order to obtain complementary features, the present invention designs a cross-modal attention fusion module (calculating the attention of different modalities separately and then fusing them). is the sequence feature representation, is the spatial feature representation, Represents the attention calculation, then the cross-modal attention fusion calculation can be expressed by the following formula: (1-8); Final pre-hospital emergency demand forecast output It can be expressed as: (1-9); in, represents the fully connected layer, It is the output of cross-modal fusion calculation.

[0036] Embodiment 2: like Figure 2As shown, this implementation provides a pre-hospital emergency demand prediction system based on multimodal fusion, including: The time series coding feature extraction module is configured to: extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; The sequence feature representation generation module is configured to: obtain external factor embedding features according to the external factor data corresponding to the pre-hospital emergency demand, and connect the external factor embedding features with the time series encoding features to obtain the sequence feature representation; The demand prediction result generation module is configured to obtain spatial feature representation based on spatial image data corresponding to the pre-hospital emergency demand, and fuse the sequence feature representation with the spatial feature representation through cross-modal attention to obtain the pre-hospital emergency demand prediction result.

[0037] The time series coding feature extraction module specifically includes: For feature extraction of time series data (including historical data on the number of emergency calls, historical population characteristics, temperature, environmental quality, etc.), the transformer time series encoding layer based on the sparse self-attention mechanism is used. It can help learn useful features in the historical stream and obtain high-level representations with dominant features.

[0038] The transformer time series encoding layer based on the sparse self-attention mechanism can be expressed as follows: (2-1); (2-2); In the formula, represents the activation function, which is Activation function, Indicates the encoder's layer, Indicates time, For community areas, represents sparse self-attention, represents a sparse matrix of the same size as the query vector, The top-u queries are restricted to keep only the query vectors with strong responses; represents a one-dimensional convolution, Indicates time Community Area The encoder The input of the layer, represents maximum pooling, is the activation function, is the key vector, Represents the extraction result of time series encoding features (if there are multiple layers, it also represents the input of the next layer i+1), represents the dimension of the key, where the query vector and key vector All passed Sure.

[0039] The sequence feature representation generation module specifically includes: Changes in pre-hospital emergency care are affected by many factors, such as frequent traffic accidents in rainy and snowy environments, which will increase the number of trauma emergency care. Therefore, external factors such as temperature, environmental quality, rainy and snowy weather, and population characteristics are also taken into consideration.

[0040] Specifically, the present invention extracts multi-dimensional continuous external features including weather conditions, air quality, temperature and population characteristics, and splices the extracted external factors into the embedding layer to obtain the external factor embedding features. , express time, represents the external characteristic length, Represents the embedding feature dimension. The external factor embedding feature is calculated through stacked convolution layers (i.e., one-dimensional convolution operation). The obtained external factor embedding feature is connected with the time series encoding feature to finally obtain the sequence feature representation .

[0041] The demand forecast result generation module specifically includes: For spatial image information, the present invention implements spatial and channel attention calculation to deal with the clustering characteristics of emergency distribution in space, and this module is lightweight. Then the spatial image features are obtained through the convolution block. , specifically, including: (2-3); (2-4); (2-5); (2-6); (2-7); in, represents one-dimensional convolution, represents a two-dimensional convolution, is the input of the attention module, represents channel attention, represents spatial attention, is the pixel value multiplied, represents the activation function, is a normalization operation, represents the average pooling operation, represents the maximum pooling operation, Indicates time, Indicates the number of channels of the feature map, and represents the height and width of the feature map, Represents the input spatial image.

[0042] The feature extraction methods based on sequence and image information are completely different, and the features obtained are also generated in different domains. In order to obtain complementary features, the present invention designs a cross-modal attention fusion module (calculating the attention of different modalities separately and then fusing them). is the sequence feature representation, is the spatial feature representation, Represents the attention calculation, then the cross-modal attention fusion calculation can be expressed by the following formula: (2-8); The final pre-hospital emergency demand forecast output can be expressed as: (2-9); in, represents the fully connected layer, It is the output of cross-modal fusion calculation.

[0043] Embodiment 3: like Figure 3 As shown, this implementation provides a computer device, which includes a processor 1001, a communication interface 1002, and a computer-readable storage medium 1003. The processor 1001, the communication interface 1002, and the computer-readable storage medium 1003 may be connected via a bus or other means.

[0044] Among them, the communication interface 1002 is used to receive and send data, the computer-readable storage medium 1003 can be stored in the memory of the electronic device, the computer-readable storage medium 1003 is used to store a computer program, the computer program includes program instructions, and the processor 1001 is used to execute the program instructions stored in the computer-readable storage medium 1003.

[0045] The processor 1001 (or CPU (Central Processing Unit)) is the computing core and control core of the electronic device, which is suitable for implementing one or more instructions, and specifically suitable for loading and executing one or more instructions to implement corresponding method processes or corresponding functions.

[0046] The processor 1001 is configured to execute the following process: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; According to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series coding features to obtain the sequence feature representation; According to the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

[0047] The specific details are described in Example 1 and will not be repeated here.

[0048] Embodiment 4: This implementation provides a computer-readable storage medium (Memory), which is a memory device in an electronic device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the electronic device and, of course, extended storage media supported by the electronic device. The computer-readable storage medium provides a storage space that stores the processing system of the electronic device.

[0049] In addition, the storage space also stores one or more instructions suitable for being loaded and executed by the processor, and these instructions may be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here may be a high-speed RAM memory, or a non-volatile memory, such as at least one disk memory; optionally, it may also be at least one computer-readable storage medium located away from the aforementioned processor.

[0050] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to implement the following process: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; According to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series coding features to obtain the sequence feature representation; According to the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

[0051] The specific details are described in Example 1 and will not be repeated here.

[0052] Embodiment 5: The present implementation provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the following process: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; According to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series coding features to obtain the sequence feature representation; According to the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

[0053] The specific details are described in Example 1 and will not be repeated here.

[0054] A person skilled in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0055] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by the computer or a data processing device such as a server, a data center, etc. that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

[0056] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting pre-hospital emergency demand based on multimodal fusion, characterized in that: The process includes: Extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; According to the external factor data corresponding to the demand for pre-hospital emergency care, the external factor embedding features are obtained, and the external factor embedding features are connected with the time series coding features to obtain the sequence feature representation; According to the spatial image data corresponding to the demand for pre-hospital emergency care, the spatial feature representation is obtained, and the sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the prediction result of the demand for pre-hospital emergency care.

2. The method for predicting pre-hospital emergency demand based on multimodal fusion according to claim 1, characterized in that: Feature extraction is performed on the historical time series data of pre-hospital emergency demand to obtain time series coding features, including: ； in, represents the activation function, Indicates the encoder's layer, Indicates time, For community areas, represents sparse self-attention, represents a one-dimensional convolution, Indicates time Community Area The encoder The input of the layer, represents maximum pooling, Represents time series encoding features.

3. The method for predicting pre-hospital emergency demand based on multimodal fusion according to claim 1, characterized in that: External factor data include: weather conditions, air quality, temperature and population size corresponding to the demand for pre-hospital emergency care in different time periods.

4. The method for predicting pre-hospital emergency demand based on multimodal fusion according to claim 1, characterized in that: According to the external factor data corresponding to the pre-hospital emergency demand, the external factor embedding features are obtained, including: The external factor data are concatenated and input into the stacked convolutional layers to obtain the external factor embedding features.

5. The method for predicting pre-hospital emergency demand based on multimodal fusion according to claim 1, characterized in that: According to the spatial image data corresponding to the pre-hospital emergency demand, the spatial feature representation is obtained ,include: ； ； ； ； ； in, represents one-dimensional convolution, represents a two-dimensional convolution, is the input of the attention module, represents channel attention, represents spatial attention, is the multiplication of pixel values, Act represents the activation function, is a normalization operation, represents the average pooling operation, represents the maximum pooling operation, Indicates time, Indicates the number of channels of the feature map, and represents the height and width of the feature map, Represents the input spatial image.

6. The method for predicting pre-hospital emergency demand based on multimodal fusion according to claim 1, characterized in that: The sequence feature representation and the spatial feature representation are fused through cross-modal attention to obtain the pre-hospital emergency demand prediction result, including: inputting the fusion result into the fully connected layer to obtain the final pre-hospital emergency demand prediction result.

7. A pre-hospital emergency demand prediction system based on multimodal fusion, characterized in that: include: The time series coding feature extraction module is configured to: extract features from the historical time series data of pre-hospital emergency demand to obtain time series coding features; The sequence feature representation generation module is configured to: obtain external factor embedding features according to the external factor data corresponding to the pre-hospital emergency demand, and connect the external factor embedding features with the time series encoding features to obtain the sequence feature representation; The demand prediction result generation module is configured to obtain spatial feature representation based on spatial image data corresponding to the pre-hospital emergency demand, and fuse the sequence feature representation with the spatial feature representation through cross-modal attention to obtain the pre-hospital emergency demand prediction result.

8. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method for predicting pre-hospital emergency demand based on multimodal fusion as described in any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for predicting pre-hospital emergency demand based on multimodal fusion as described in any one of claims 1-6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, and when the computer program is executed by a processor, it implements the method for predicting pre-hospital emergency demand based on multimodal fusion as described in any one of claims 1 to 6.

Images