First-aid material recommendation method and system based on AI

Abstract

The invention provides an AI-based first-aid material recommendation method and system. The system obtains real-time data and physiological state information of a current environment of a user, carries out integration processing, and generates environment and health condition data of the user. The relevance between environmental factors and health conditions is analyzed by using a context awareness algorithm, and the emergency situation type is determined by combining a multi-modal fusion model. According to the determined emergency situation type, the system combines a first-aid material knowledge graph constructed by a graph neural network, uses a deep learning algorithm to generate a first-aid material list, and comprehensively considers the effectiveness and portability of materials and specific requirements of a user. And the system performs priority ranking on the materials according to the importance and availability of the articles, and displays the materials to the user by adopting an augmented reality technology. Meanwhile, the navigation information of the nearest first-aid material storage position is adjusted based on the real-time traffic data, and the user is helped to quickly find the required materials. According to the technical scheme provided by the invention, the first-aid material recommendation efficiency is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of intelligent first aid technology, and in particular, to an AI-based first aid material recommendation method and system. Background Art

[0002] In emergency situations, such as natural disasters, sudden diseases, or accidents, quickly obtaining appropriate first aid materials is crucial for saving lives and reducing injuries. Users need an intelligent system to real-time evaluate their current environment and health status and provide personalized first aid material recommendations. Such a system must be able to integrate multiple data sources, including the user's physiological state information and real-time data of the surrounding environment, to ensure that the recommended materials are both effective and portable, while meeting the specific needs of the user.

[0003] Currently, most first aid material recommendation systems mainly rely on preset standardized lists or simple user inputs, lacking dynamic evaluation of the user's specific environment and health status. Some relatively advanced systems have begun to use sensors and mobile devices to collect the user's physiological data, but these systems usually cannot comprehensively analyze the complex correlation between environmental factors and personal health status, resulting in inaccurate and non-personalized recommendation results.

[0004] However, most existing systems have not fully utilized context-aware algorithms to analyze the correlation between the user's surrounding environmental factors and personal health status, thus affecting the accuracy and applicability of the recommendations. Traditional recommendation methods have not combined advanced AI technologies such as deep learning and graph neural networks, making it difficult to generate personalized recommendations that comprehensively consider the effectiveness, portability of materials, and the specific needs of users. The navigation and display links after material recommendation in existing systems are relatively simple, and they have not fully utilized augmented reality technology and real-time traffic data for dynamic adjustment, reducing the user experience and rescue efficiency. Summary of the Invention

[0005] The embodiments of the present application provide an AI-based first aid material recommendation method and system to solve the problem of poor first aid material recommendation effect in the prior art.

[0006] In a first aspect, the embodiments of the present application provide an AI-based first aid material recommendation method, including:

[0007] Obtain the real-time data of the user's current environment and the user's physiological state information, integrate and process the real-time data and the user's physiological state information to obtain the integrated user environment and health status data;

[0008] Based on the integrated user environment and health status data, analyze the correlation between the user's surrounding environmental factors and personal health status through a context-aware algorithm, and then use a pre-trained multi-modal fusion model to determine the type of emergency situation the user encounters;

[0009] According to the type of emergency, combined with the first-aid supply knowledge graph constructed based on the graph neural network, use deep learning algorithms to perform intelligent recommendation processing on the first-aid supply information in the knowledge graph, generate a corresponding first-aid supply list, where the recommendation process comprehensively considers the effectiveness, portability of the supplies, and the specific needs of the user;

[0010] Perform a priority sorting process on the first-aid supply list according to the importance and availability of the items, obtain a first-aid supply list arranged by priority, and use augmented reality technology to display it to the user through the user interface, and at the same time adjust the navigation information of the nearest first-aid supply storage location based on real-time traffic data.

[0011] Optionally, based on the integrated user environment and health status data, analyze the correlation between the user's surrounding environmental factors and personal health status through a context-aware algorithm, and then use a pre-trained multi-modal fusion model to determine the type of emergency encountered by the user, including:

[0012] Based on the integrated user environment and health status data, use an improved context-aware algorithm combined with a Bayesian network to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure, and obtain the specific context mode and potential change trend where the user is currently located;

[0013] According to the specific context mode and potential change trend, input it into a context-health impact factor prediction model optimized based on transfer learning, evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate a health risk score;

[0014] Based on the health risk score, use a reinforcement learning strategy to simulate the best response measures in different emergencies, and combine the statistics of the occurrence frequency and severity of emergencies in the historical case library, and use a pre-trained and adaptively adjusted multi-modal fusion model for further refined analysis to determine the specific type of emergency that the user may encounter and its urgency.

[0015] Optionally, based on the integrated user environment and health status data, use an improved context-aware algorithm combined with a Bayesian network to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure, and obtain the specific context mode and potential change trend where the user is currently located, including:

[0016] Based on the integrated user environment and health status data, use an improved context-aware algorithm to identify key environmental factors and physiological indicators;

[0017] Based on the described key environmental factors and physiological indicators, perform probabilistic inference processing through the constructed Bayesian network model to quantify the interactions between these factors and their impacts on the user's health status;

[0018] Based on the probability distribution output by the Bayesian network model, combine time series analysis methods to predict the changing trends of the user's environment and health status over a period of time in the future, and generate a changing trend prediction result;

[0019] Utilize the changing trend prediction result, through pattern recognition processing of the probability distribution and the changing trend, determine the specific situation pattern in which the user is currently located, and generate the future development direction of the specific situation pattern.

[0020] Optionally, input the specific situation pattern and potential changing trend into a situation-health impact factor prediction model optimized based on transfer learning to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate a health risk score, including:

[0021] According to the specific situation pattern and potential changing trend, construct a data set containing the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure;

[0022] Based on the constructed data set, input it into a situation-health impact factor prediction model optimized based on transfer learning for processing to obtain a preliminary health impact analysis result;

[0023] Utilize the preliminary health impact analysis result to further analyze and process the immediate and long-term health impacts under different combinations of environmental factors and physiological states, and generate a detailed health impact assessment;

[0024] According to the detailed health impact assessment, calculate and generate a health risk score corresponding to each combination of environmental factors and physiological states.

[0025] Optionally, based on the health risk score, use reinforcement learning strategies to simulate the best response measures in different emergency situations, and combine the statistics of the occurrence frequency and severity of emergency situations in the historical case library, and use a pre-trained and adaptively adjusted multi-modal fusion model for further refined analysis to determine the specific types of emergency situations that the user may encounter and their urgency levels, including:

[0026] Based on the health risk score, use reinforcement learning algorithms to simulate various response measures taken in different emergency situations to obtain the best response measure set for each emergency situation;

[0027] According to the set of optimal response measures, combined with the statistical data of the occurrence frequency and severity of emergencies recorded in the historical case database, weight adjustment processing is performed on the set of optimal response measures to obtain an adjusted set of optimal response measures;

[0028] Based on the adjusted set of optimal response measures and relevant historical data, a pre-trained and adaptively adjusted multi-modal fusion model is used for further analysis and processing to generate a prediction result;

[0029] According to the prediction result, determine the specific type of emergency encountered by the user, and determine its urgency by evaluating the occurrence probability and potential impact of each emergency.

[0030] Optionally, according to the type of emergency, combined with the first-aid supplies knowledge graph constructed based on a graph neural network, a deep learning algorithm is used to perform intelligent recommendation processing on the first-aid supplies information in the knowledge graph to generate a corresponding list of first-aid supplies, where the recommendation process comprehensively considers the effectiveness, portability of the supplies, and the specific needs of the user, including:

[0031] According to the type of emergency, extract the first-aid supplies nodes related to the emergency from the first-aid supplies knowledge graph constructed based on a graph neural network;

[0032] Using a deep learning algorithm, analyze and process the first-aid supplies nodes to obtain the evaluation results of the effectiveness and portability of each supply node relative to the user's needs in the current emergency;

[0033] Based on the evaluation results and combined with the specific needs of the user, perform priority sorting processing on the first-aid supplies nodes to generate a sorted list of first-aid supplies;

[0034] According to the sorted list of first-aid supplies, generate a list of first-aid supplies that meet the current emergency and meet the specific needs of the user.

[0035] Optionally, perform priority sorting processing on the list of first-aid supplies according to the importance and availability of the items to obtain a list of first-aid supplies arranged in priority, and use augmented reality technology to display it to the user through the user interface, and at the same time adjust the navigation information of the nearest first-aid supplies storage location based on real-time traffic data, including:

[0036] Based on the importance of each item in the first-aid supplies list and its availability in the current environment, evaluate each supply in the list to generate the comprehensive priority of each supply;

[0037] According to the comprehensive priority of each supply, sort the first-aid supplies list to obtain a list of first-aid supplies arranged in priority;

[0038] Using augmented reality technology, the prioritized list of first-aid supplies is visually presented to the user through the user interface;

[0039] Based on the user's current location, combining the prioritized list of first-aid supplies and real-time traffic data, determine the best route to the nearest first-aid supply storage point and provide dynamically updated navigation information.

[0040] In a second aspect, an AI-based first-aid supply recommendation system provided by an embodiment of the present application includes:

[0041] An acquisition and integration module for acquiring real-time data of the user's current environment and the user's physiological state information, integrating and processing the real-time data and the user's physiological state information to obtain integrated user environment and health status data;

[0042] An analysis and determination module for analyzing the correlation between the surrounding environmental factors of the user and the personal health status through a context awareness algorithm based on the integrated user environment and health status data, and then using a pre-trained multi-modal fusion model to determine the type of emergency encountered by the user;

[0043] A processing and generation module for, according to the type of emergency, combining a first-aid supply knowledge graph constructed based on a graph neural network, and using a deep learning algorithm to perform intelligent recommendation processing on the first-aid supply information in the knowledge graph to generate a corresponding list of first-aid supplies, where the recommendation process comprehensively considers the effectiveness, portability of the supplies, and the specific needs of the user;

[0044] A sorting and navigation module for performing priority sorting on the list of first-aid supplies according to the importance and availability of the items to obtain a prioritized list of first-aid supplies, and presenting it to the user through the user interface using augmented reality technology, and at the same time adjusting the navigation information of the nearest first-aid supply storage location based on real-time traffic data.

[0045] In a third aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an AI-based first-aid supply recommendation method as described in any item of the first aspect.

[0046] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it implements an AI-based first-aid supply recommendation method as described in any item of the first aspect.

[0047] In the embodiments of the present application, real-time data of the user's current environment and the user's physiological state information are obtained, and the real-time data and the user's physiological state information are integrated and processed to obtain integrated user environment and health status data; based on the integrated user environment and health status data, the relevance between the environmental factors around the user and the personal health status is analyzed through a context-aware algorithm, and then the type of emergency encountered by the user is determined by using a pre-trained multi-modal fusion model; according to the type of emergency, combined with an emergency supplies knowledge graph constructed based on a graph neural network, a deep learning algorithm is used to perform intelligent recommendation processing on the emergency supplies information in the knowledge graph to generate a corresponding emergency supplies list, where the recommendation process comprehensively considers the effectiveness, portability, and specific needs of the user; the emergency supplies list is sorted according to the importance and availability of the items to obtain an emergency supplies list arranged in priority, and an augmented reality technology is used to display it to the user through a user interface, and at the same time, the navigation information of the nearest emergency supplies storage location is adjusted based on real-time traffic data.

[0048] Through real-time data acquisition and intelligent analysis, the system can determine the type of emergency faced by the user in a short time and provide accurate material recommendations, shortening the emergency response time. The recommendation process comprehensively considers the effectiveness, portability, and specific needs of the user, ensuring that the provided materials are not only suitable for the current emergency but also meet the personalized needs of the user. Sorting by the importance and availability of the items enables the user to obtain the most urgently needed materials first, improving the resource utilization efficiency. Using augmented reality technology to display the recommendation results provides an intuitive and friendly user interface, facilitating user understanding and operation. Adjusting the navigation information of the nearest emergency supplies storage location based on real-time traffic data ensures that the user can reach the supplies storage point most quickly, further enhancing the rescue speed and effectiveness.

[0049] Furthermore, in the embodiments of the present application, the context-aware algorithm is used to analyze the user environment and health status data, and combined with a pre-trained multi-modal fusion model to determine the type of emergency encountered by the user. An improved context-aware algorithm combined with a Bayesian network is used to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure to obtain the user's current specific context pattern and potential change trend. According to the specific context pattern and potential change trend, it is input into a context-health impact factor prediction model optimized based on transfer learning to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate a health risk score.

[0050] Based on the health risk score, use reinforcement learning strategies to simulate the best response measures in different emergency situations, and combine the statistics of the occurrence frequency and severity of emergency situations in the historical case database. Use a pre-trained and adaptively adjusted multi-modal fusion model for further refined analysis to determine the specific types of emergency situations that the user may encounter and their urgency levels. According to the determined types of emergency situations, the system combines the first-aid supply knowledge graph constructed by a graph neural network and uses a deep learning algorithm to generate a personalized first-aid supply list. According to the type of emergency situation, extract the first-aid supply nodes related to the emergency situation from the first-aid supply knowledge graph constructed by the graph neural network. Use a deep learning algorithm to analyze and process the extracted first-aid supply nodes to obtain the evaluation results of the effectiveness and portability of each supply node relative to the user's needs in the current emergency situation. Based on the above evaluation results and combined with the specific needs of the user, perform a priority ranking process on the first-aid supply nodes to generate a sorted first-aid supply list. According to the sorted first-aid supply list, generate a first-aid supply list that meets the current emergency situation and conforms to the specific needs of the user.

[0051] By using intelligent means to quickly determine the type of emergency situation and provide accurate supply recommendations, the emergency response time is significantly shortened. The recommendation process comprehensively considers the effectiveness, portability, and specific needs of the user, ensuring that the provided supplies are not only suitable for the current emergency situation but also meet the user's personalized needs. Priority ranking is carried out according to the importance and availability of the items, enabling the user to obtain the most urgently needed supplies first and improving the resource utilization efficiency. The use of augmented reality technology to display the recommendation results and adjust the navigation information based on real-time traffic data provides an intuitive and user-friendly interface, facilitating user understanding and operation. The dynamic navigation support helps the user quickly find the nearest first-aid supply storage point, further enhancing the rescue speed and effectiveness.

[0052] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a flowchart of a first-aid supply recommendation method based on AI provided by an embodiment of the present application;

[0055] Figure 2Schematic structural diagram of a first-aid supplies recommendation system based on AI provided by an embodiment of the present application;

[0056] Figure 3 Schematic structural diagram of a computing device provided by an embodiment of the present application. Detailed implementation manners

[0057] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application.

[0058] In some processes described in the specification, claims and above-mentioned drawings of the present application, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0059] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present application.

[0060] Figure 1 Flowchart of a first-aid supplies recommendation method based on AI provided by an embodiment of the present application, as Figure 1 shown, the method includes:

[0061] 101. Obtain real-time data of the user's current environment and user physiological state information, integrate and process the real-time data and user physiological state information to obtain integrated user environment and health condition data;

[0062] In this step, it involves collecting and integrating various types of data, including the user's real-time location, environmental parameters, and physiological state information. These data are used to construct a comprehensive dataset of the user's environment and health status. Environmental data can be obtained through sensor networks or sensors on mobile devices, while physiological state information is collected through devices such as wearable medical devices or smartwatches. The integration process refers to uniformly formatting these scattered data sources and performing preliminary analysis to ensure the consistency and accuracy of subsequent processing.

[0063] First, the system collects real-time data from multiple data sources, including the GPS coordinates of the user's location, meteorological data provided by environmental monitoring stations, and physiological indicators such as heart rate and blood pressure uploaded by the user's wearable devices. Then, this data is transmitted to the cloud server for integration processing. The server uses preprocessing algorithms to clean, standardize, and extract features from the data, and finally generates a comprehensive dataset containing the user's current location, environmental conditions, and personal health status. This dataset provides a solid foundation for subsequent context-aware analysis.

[0064] In the example of this application, in an urban first-aid scenario, when the user triggers an emergency call, the system immediately starts the data collection process. Suppose the user encounters high-temperature weather outdoors and feels unwell, and their smartphone and smart bracelet will automatically send real-time location information, surrounding environmental temperature and humidity data, and physiological state information such as heart rate and blood pressure to the first-aid platform. After receiving this data, the first-aid platform integrates and processes it to generate a complete dataset of the user's environment and health status, preparing for further analysis.

[0065] 102. Based on the integrated user environment and health status data, analyze the correlation between the user's surrounding environmental factors and personal health status through context-aware algorithms, and then use a pre-trained multi-modal fusion model to determine the type of emergency situation the user has encountered;

[0066] In this step, use context-aware algorithms combined with Bayesian networks to dynamically model and comprehensively analyze the user's specific context patterns and their potential change trends. Context-aware algorithms can identify the complex relationship between the environmental characteristics where the user is currently located and their health status, while the multi-modal fusion model uses machine learning methods to fuse data from different sources, thereby accurately predicting the type of emergency situation the user may face. This analysis not only considers immediate risks but also evaluates long-term impacts, providing a scientific basis for subsequent recommendations.

[0067] The system inputs the integrated user environment and health condition data into an improved context awareness algorithm. This algorithm combines a Bayesian network to dynamically model and comprehensively analyze the data, and obtains the user's current specific context pattern and potential change trend. Then, these results are fed into a pre-trained multi-modal fusion model, which further refines the analysis to determine the specific types of emergency situations that the user may encounter and their urgency. During this process, the model also refers to the statistics of the occurrence frequency and severity of similar events in the historical case library to ensure the accuracy and reliability of the prediction results.

[0068] In the example of this application, continuing the above-mentioned urban first aid scenario, after the first aid platform receives the integrated data set of the user, it starts to analyze the specific context. The context awareness algorithm finds that the user is in a high-temperature environment, with an abnormally elevated heart rate and significant blood pressure fluctuations. Combining historical data, the system infers that the user may have a relatively high risk of heat stroke. Subsequently, based on this information, the multi-modal fusion model further confirms that the user is indeed in a high-risk situation and determines that cooling measures need to be taken immediately. The system thus infers that the specific type of emergency situation is "heat stroke", laying the foundation for the subsequent material recommendation.

[0069] Optionally, in step 102, based on the integrated user environment and health condition data, the relevance between the environmental factors around the user and the personal health condition is analyzed through a context awareness algorithm, and then the type of emergency situation encountered by the user is determined using a pre-trained multi-modal fusion model, including: based on the integrated user environment and health condition data, an improved context awareness algorithm combined with a Bayesian network is used to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure, to obtain the user's current specific context pattern and potential change trend; according to the specific context pattern and potential change trend, it is input into a context-health impact factor prediction model optimized based on transfer learning to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate a health risk score; based on the health risk score, a reinforcement learning strategy is used to simulate the best response measures in different emergency situations, and combined with the statistics of the occurrence frequency and severity of emergency situations in the historical case library, a pre-trained and adaptively adjusted multi-modal fusion model is used for further refined analysis to determine the specific type of emergency situation that the user may encounter and its urgency.

[0070] Optionally, based on the integrated user environment and health status data in step 102, an improved context awareness algorithm is used in combination with a Bayesian network to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure, to obtain the specific context mode and potential change trend in which the user is currently located, including: based on the integrated user environment and health status data, using an improved context awareness algorithm to identify key environmental factors and physiological indicators; according to the key environmental factors and physiological indicators, performing probabilistic inference processing through the constructed Bayesian network model to quantify the interaction between these factors and their impact on the user's health status; based on the probability distribution output by the Bayesian network model, combining time series analysis methods to predict the change trend of the user's environment and health status in a future period of time, generating a change trend prediction result; using the change trend prediction result, through pattern recognition processing of the probability distribution and change trend, determining the specific context mode in which the user is currently located, and generating the future development direction of the specific context mode.

[0071] In this step, the context awareness algorithm is an intelligent algorithm that can identify and understand the environment in which the user is located and its changes. It not only considers static environmental factors but also dynamically monitors the user's physiological state. Through the improved context awareness algorithm, the system can more accurately identify key environmental factors and physiological indicators, providing a basis for subsequent analysis. The Bayesian network model is a probabilistic graphical model used to represent the conditional dependence relationships between variables. In this solution, the Bayesian network is used to quantify the interaction between environmental factors and physiological indicators and their impact on the user's health status. Through probabilistic inference processing of these variables, the change of the user's health risk under different combinations can be evaluated. The time series analysis method is used to predict the change trend of the user's environment and health status in a future period of time. By analyzing the time series characteristics of historical data, the system can estimate the future development direction to help better understand and respond to potential risks. Pattern recognition processing refers to using machine learning techniques to identify and classify different probability distributions and change trend patterns. Through the recognition of these patterns, the system can determine the specific context mode of the user currently and predict its future development direction.

[0072] First, the system uses an improved context awareness algorithm to identify key environmental factors and physiological indicators that have a significant impact on the user's health from the integrated user environment and health status data.

[0073] Secondly, according to the identified key factors and indicators, the system constructs a Bayesian network model and performs probabilistic inference processing. This model quantifies the interaction between these factors and their impact on the user's health status, generating a corresponding probability distribution.

[0074] Next, based on the probability distribution output by the Bayesian network model and combined with the time series analysis method, the system predicts the changing trends of the user's environment and health status over a period of time in the future, and generates a prediction result of the changing trends.

[0075] Finally, the system uses the prediction result of the changing trends, and through pattern recognition processing of the probability distribution and the changing trends, determines the specific situation pattern in which the user is currently located, and predicts the future development direction of this pattern.

[0076] In the embodiment of the present application, in an outdoor sports first aid scenario, it is assumed that an athlete feels unwell during a long-distance running training in a high-temperature environment and triggers an emergency call. At this time, the system starts the following process:

[0077] The system identifies key environmental factors (such as the current temperature of 35°C and humidity of 70%) and physiological indicators (such as a heart rate reaching 180 beats per minute and elevated blood pressure) from the integrated user environment and health status data. These information indicate that the user is in a high-risk environment and has an abnormal physiological state.

[0078] The system constructs a Bayesian network model based on the identified key factors and indicators, and performs probability inference processing. The model quantifies the interaction between high temperature, high humidity and rapid heart rate, high blood pressure, and concludes that the user has a relatively high risk of heat stroke, and generates a corresponding probability distribution.

[0079] Based on the probability distribution output by the Bayesian network model, the system combines the time series analysis method to predict that the user's body temperature will continue to rise and the heart rate and blood pressure will further increase in the future, which may lead to severe dehydration or heat stroke.

[0080] Finally, the system uses the prediction result of the changing trends, and through pattern recognition processing of the probability distribution and the changing trends, determines that the user is currently in the specific situation pattern of "heat stroke", and predicts that this pattern will continue to deteriorate. If no measures are taken in time, it may develop into severe heat stroke.

[0081] Through this series of steps, the system not only accurately identifies the type of the user's current emergency situation, but also predicts the future development direction, providing a scientific basis for subsequent first aid material recommendation and response measures. This method based on the improved context awareness algorithm and Bayesian network significantly improves the accuracy and predictability of emergency situation judgment, and provides more personalized and effective first aid support for users.

[0082] Optionally, the input in step 102 according to the specific scenario mode and potential change trend is input into the scenario-health impact factor prediction model optimized based on transfer learning to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate health risk scores, including: constructing a data set containing the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure according to the specific scenario mode and potential change trend; based on the constructed data set, input it into the scenario-health impact factor prediction model optimized based on transfer learning for processing to obtain a preliminary health impact analysis result; using the preliminary health impact analysis result, further analyze and process the immediate and long-term health impacts under different combinations of environmental factors and physiological states to generate a detailed health impact assessment; according to the detailed health impact assessment, calculate and generate the health risk score corresponding to each combination of environmental factors and physiological states.

[0083] Optionally, based on the health risk score, use a reinforcement learning strategy to simulate the best response measures in different emergency situations, and combine the statistics of the occurrence frequency and severity of emergency situations in the historical case library, and use a pre-trained and adaptively adjusted multi-modal fusion model for further refined analysis to determine the specific types of emergency situations that the user may encounter and their urgency, including: based on the health risk score, use a reinforcement learning algorithm to simulate various response measures taken in different emergency situations to obtain the best response measure set for each emergency situation; according to the best response measure set, combine the statistics of the occurrence frequency and severity of emergency situations recorded in the historical case library to adjust the weights of the best response measure set to obtain an adjusted best response measure set; based on the adjusted best response measure set and related historical data, use a pre-trained and adaptively adjusted multi-modal fusion model for further analysis and processing to generate a prediction result; according to the prediction result, determine the specific type of emergency situation that the user encounters, and determine its urgency by evaluating the occurrence probability and potential impact of each emergency situation.

[0084] In this step, the context-health impact factor prediction model optimized by transfer learning is a machine learning model improved by using transfer learning technology, aiming to enhance the generalization ability of the model in new contexts. This model can quickly adapt to new data distributions and accurately evaluate the immediate and long-term impacts of different combinations of environmental factors and physiological states on the user's health. The health risk score is a quantitative assessment of the health risks under different combinations of environmental factors and physiological states. A higher score indicates a greater health risk, helping the system prioritize the handling of high-risk situations. The reinforcement learning strategy is a machine learning method that finds the best coping strategies by simulating the effects of different coping measures. It combines the statistics of the occurrence frequency and severity of emergency situations in the historical case library to ensure that the recommended coping measures are both effective and reasonable. The multi-modal fusion model is a model that comprehensively considers multiple data sources for further refined analysis to improve the accuracy of prediction.

[0085] First, according to the specific context pattern and potential change trends, a data set is constructed that includes the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure. The constructed data set is input into the context-health impact factor prediction model optimized by transfer learning for processing to obtain the preliminary health impact analysis results.

[0086] Secondly, using the preliminary health impact analysis results, further analysis and processing are carried out on the immediate and long-term health impacts under different combinations of environmental factors and physiological states to generate a detailed health impact assessment. According to the detailed health impact assessment, the health risk score corresponding to each combination of environmental factors and physiological states is calculated and generated.

[0087] Then, based on the health risk score, various coping measures taken in different emergency situations are simulated using the reinforcement learning algorithm to obtain the best coping measure set for each emergency situation. According to the best coping measure set, combined with the statistics of the occurrence frequency and severity of emergency situations recorded in the historical case library, weight adjustment processing is performed on the best coping measure set to obtain the adjusted best coping measure set.

[0088] Finally, based on the adjusted best coping measure set and relevant historical data, further analysis and processing are carried out using a pre-trained and adaptively adjusted multi-modal fusion model to generate a prediction result. According to the prediction result, the specific type of emergency situation encountered by the user is determined, and its urgency is determined by evaluating the occurrence probability and potential impact of each emergency situation.

[0089] In the embodiment of this application, in an urban first-aid scenario, assume that a user feels unwell in a high-temperature environment and triggers an emergency call. The system starts the following process:

[0090] First, the system constructs a detailed dataset based on the user's real-time location information (such as GPS coordinates), environmental temperature and humidity (such as 35°C, 70%), and physiological state information such as heart rate (such as 180 beats per minute) and blood pressure (such as 160 / 90 mmHg).

[0091] The dataset is input into the context-health impact factor prediction model optimized by transfer learning. The model improves its adaptability to new environmental data through transfer learning, generates preliminary health impact analysis results, indicating that the user has a relatively high risk of heat stroke.

[0092] Secondly, the system uses the preliminary health impact analysis results to conduct an in-depth analysis of the immediate and long-term health impacts under the combination of rapid heart rate and high blood pressure in high-temperature and high-humidity environments, and generates a detailed health impact assessment.

[0093] Based on the detailed health impact assessment, the system calculates and generates a health risk score corresponding to each combination of environmental factors and physiological states. For example, the health risk score in a high-temperature and high-humidity environment is 90 points (out of 100), indicating an extremely high risk.

[0094] Then, based on the health risk scores, the system uses a reinforcement learning algorithm to simulate response measures in different emergency situations, such as immediate cooling, replenishing water, and finding a shady place to rest.

[0095] Combined with the statistical data on the occurrence frequency and severity of similar events in the historical case library, the system adjusts the weights of these response measures and obtains an adjusted set of optimal response measures, such as immediate cooling (weight 0.8), replenishing water (weight 0.7), and finding a shady place to rest (weight 0.6).

[0096] Finally, based on the adjusted set of optimal response measures and relevant historical data, the system uses a pre-trained and adaptively adjusted multi-modal fusion model for further analysis and processing to generate prediction results. According to the prediction results, the system determines that the specific type of emergency situation the user has encountered is "heat stroke", and determines its urgency level as "very high" by evaluating the occurrence probability and potential impact of each emergency situation.

[0097] Through this series of steps, the system not only accurately assesses the user's health risks, but also provides scientific and reasonable suggestions for response measures, significantly improving the efficiency and success rate of first aid. This method of context-health impact factor prediction model optimized by transfer learning and reinforcement learning strategy ensures that the system can provide personalized and efficient first aid support in complex and changing environments.

[0098] This application takes into account that in the fields of first aid and health management, accurately assessing the immediate and long-term health risks of users is crucial for taking effective measures in a timely manner. Traditional health risk assessment methods often rely on static data or preset rule sets and are difficult to adapt to complex and changing real-world environments. To improve the accuracy and flexibility of assessment, researchers have introduced a context-health impact factor prediction model optimized by transfer learning and developed a new set of formulas to quantify health risks under different combinations of environmental factors and physiological states.

[0099] This set of formulas aims to process environmental factors (such as temperature, humidity) and physiological states (such as heart rate, blood pressure) through non-linear transformation, combine weight coefficients and bias terms, and generate a comprehensive health risk score. This score can not only reflect the health risks in the current context but also predict future health change trends through further analysis, providing a scientific basis for personalized medicine and emergency rescue. Therefore, a new alternative is proposed, which includes:

[0100] Optionally, the input in step 102 according to the specific context pattern and potential change trend into the context-health impact factor prediction model optimized by transfer learning to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states and generate a health risk score, including:

[0101] Based on the specific context pattern and potential change trend, construct a data set containing the user's real-time location information, environmental factors such as environmental temperature and humidity, and physiological state information such as heart rate and blood pressure;

[0102] Using the data set, through the non-linear function and respectively transform and process the environmental factor E and the physiological state P, where α 1 is a parameter controlling the slope of the curve, μ E is the reference value of the environmental factor, μ P is the reference value of the physiological state, and σ P is the standard deviation;

[0103] Based on the transformed environmental factors and physiological states, combine the weight coefficients w 1 and w 2 and the bias term b to calculate the health risk score R;

[0104] Calculate the health risk score R through the following formula:

[0105] R = w 1 ·f(E) + w 2 ·g(P) + b

[0106] where w 1 is the influence weight of the environmental factor E on the health risk; w2 w is the influence weight of environmental factors on health risks; b is a bias term used to adjust the model output;

[0107] Input the health risk score R into the context-health impact factor prediction model optimized by transfer learning for further analysis and processing to obtain a detailed health risk assessment result.

[0108] The following is a detailed annotation description of each parameter of this formula:

[0109] In the formula R = w 1 ·f(E) + w 2 ·g(P) + b, R is the health risk score, indicating the degree of health risk of the user in the current context; w 1 is the influence weight of environmental factors on health risks; w 2 is the influence weight of physiological state on health risks; f(E) is the environmental factor score after conversion, calculated and output through a non-linear function ; g(P) is the physiological state score after conversion, calculated through the Gaussian function ; b is a bias term used to adjust the baseline level of the model output;

[0110] The following is a brief introduction to the design reasons for each item of this formula:

[0111] In the formula for calculating the health risk score, the design reason for w 1 ·f(E) is that environmental factors (such as temperature, humidity) have a significant impact on health, but this impact is not linear. By converting environmental factors through the Sigmoid function f(E) and multiplying by the weight w 1 , the contribution of environmental factors to health risks can be quantified. The weight w 1 ensures that the influence of environmental factors accounts for an appropriate proportion in the total score; the design reason for w 2 ·g(P) is that physiological state (such as heart rate, blood pressure) directly reflects the user's immediate health condition. By converting physiological state through the Gaussian function g(P) and multiplying by the weight w 2 , the contribution of physiological state to health risks can be quantified. The weight w 2 ensures that the influence of physiological state accounts for an appropriate proportion in the total score; the bias term b provides a basic health risk score, so that the system can give a reasonable initial score even when all input features are zero. This helps to avoid the problem of too low a score in extreme cases.

[0112] The following is the reason for adding each item:

[0113] Adding each item (w 1 ·f(E) and w 2Adding (·g(P)) and the bias term b is to obtain a comprehensive health risk score R. The addition operation enables the effects of various factors to be accumulated together to form an overall score. The addition of the bias term b ensures the baseline level of the score, preventing the score from being too low or too high.

[0114] In this way, the formula not only considers the individual effects of environmental factors and physiological states, but also makes a comprehensive adjustment through the weight coefficients and the bias term, ensuring the comprehensiveness and accuracy of the health risk score. This method is particularly applicable to first-aid scenarios where it is necessary to quickly assess health risks and take corresponding measures, improving the practicality and reliability of the system.

[0115] Suppose in a high-temperature environment, an athlete feels unwell while running outdoors and triggers an emergency call. The system needs to evaluate the immediate and long-term health risks he faces and generate a corresponding health risk score. The following are the specific steps and calculation processes:

[0116] The following is data collection and preparation:

[0117] Real-time location information: The user is located in a city park (GPS coordinates are known).

[0118] The following are environmental factors:

[0119] Temperature E temp = 35°C

[0120] Humidity E humidity = 70%

[0121] The following are physiological states:

[0122] Heart rate P HR = 180 beats per minute

[0123] Blood pressure P BP = 160 / 90 mmHg

[0124] The following is environmental factor conversion:

[0125] Using a non-linear function where α 1 = 1, μ E = 25°C (reference temperature).

[0126] Calculate the converted environmental factor values f(E temp ) and f(E humidity ):

[0127]

[0128] Calculate the converted physiological state values g(P HR ) and g(P BP ):

[0129]

[0130] The following are the weight coefficients and bias terms: w 1 = 0.7 (weight of environmental factors); w 2 = 0.3 (weight of physiological state); b = 0.1 (bias term);

[0131] Calculate the health risk score R:

[0132] R = W 1 ·f(E) + W 2 ·g(P) + b

[0133] R = 0.7·(0.999 + 1.0) + 0.3·(0.011 + 0.135) + 0.1

[0134] R = 0.7·1.999 + 0.3·0.146 + 0.1

[0135] R ≈ 1.3993 + 0.0438 + 0.1 = 1.5431

[0136] According to the above calculation, the health risk score is 1.5431, indicating that the user faces a relatively high health risk. In a high-temperature environment, the user's heart rate has increased abnormally (180 beats per minute), and the blood pressure is also at a relatively high level (160 / 90 mmHg). These physiological indicators suggest that the user may already be in the early stage of heat stroke. Continuing to be exposed to a high-temperature and high-humidity environment, if cooling measures are not taken in time, it may lead to severe dehydration or heat stroke, threatening life safety. The system should immediately recommend that the user take cooling measures, such as using ice packs, finding a shady place to rest, and replenishing water. At the same time, it is recommended to contact professional medical personnel as soon as possible for further examination and treatment.

[0137] Through the application of this context-health impact factor prediction model optimized based on transfer learning and the non-linear transformation formula, the system can more accurately evaluate the user's health risk, provide timely and effective first aid guidance for the user, and significantly improve the first aid efficiency and success rate. This method is particularly applicable to emergency rescue scenarios in extreme environments such as high temperatures, ensuring that users can obtain appropriate first aid support in the shortest possible time.

[0138] 103. According to the type of emergency situation, in combination with the first aid material knowledge graph constructed based on the graph neural network, use the deep learning algorithm to perform intelligent recommendation processing on the first aid material information in the knowledge graph, generate the corresponding first aid material list, where the recommendation process comprehensively considers the effectiveness, portability of the materials, and the specific needs of the user;

[0139] In this step, it aims to extract relevant nodes from the first-aid supply knowledge graph constructed by the graph neural network according to the determined type of emergency situation, and use deep learning algorithms for intelligent recommendation. The knowledge graph contains a large amount of information about first-aid supplies, including attributes such as the types, uses, effectiveness, and portability of the supplies. The recommendation process not only considers the basic characteristics of the supplies but also combines the specific needs of the user, such as allergy history, special medical conditions, etc., to ensure that the recommended supplies are both practical and personalized.

[0140] The system extracts relevant first-aid supply nodes from the first-aid supply knowledge graph constructed by the graph neural network according to the determined type of emergency situation. Then, it uses deep learning algorithms to analyze and process these nodes, evaluating the effectiveness and portability of each supply node relative to the user's needs in the current emergency situation. Based on these evaluation results, the system further combines the specific needs of the user, such as drug allergy history or special medical conditions, to prioritize the supply nodes and generate a sorted list of first-aid supplies. Finally, according to the sorting results, the system generates a list of first-aid supplies that meet the current emergency situation and the specific needs of the user.

[0141] In the example of this application, after determining that the user is facing an emergency of "heat stroke", the system extracts supply nodes related to heat stroke first aid from the first-aid supply knowledge graph, such as ice packs, essential balm, electrolyte drinks, etc. The deep learning algorithm evaluates the effectiveness and portability of these supplies. Considering that the user has no allergy history but has a history of hypertension, the system gives priority to recommending non-irritating electrolyte drinks and excludes items containing alcohol. The finally generated list of first-aid supplies includes ice packs, essential balm, and electrolyte drinks, ensuring that the user can quickly obtain appropriate first-aid supplies in a high-temperature environment.

[0142] Optionally, in step 103, according to the type of emergency situation, combined with the first-aid supply knowledge graph constructed based on the graph neural network, use deep learning algorithms to perform intelligent recommendation processing on the first-aid supply information in the knowledge graph, and generate a corresponding list of first-aid supplies. The recommendation process comprehensively considers the effectiveness, portability of the supplies, and the specific needs of the user, including: extracting first-aid supply nodes related to the emergency situation from the first-aid supply knowledge graph constructed based on the graph neural network according to the type of emergency situation; using deep learning algorithms to analyze and process the first-aid supply nodes to obtain the evaluation results of the effectiveness and portability of each supply node relative to the user's needs in the current emergency situation; based on the evaluation results, and combined with the specific needs of the user, perform a priority sorting process on the first-aid supply nodes to generate a sorted list of first-aid supplies; according to the sorted list of first-aid supplies, generate a list of first-aid supplies that meet the current emergency situation and the specific needs of the user.

[0143] In this step, the first-aid supply knowledge graph constructed by the graph neural network is a structured data representation. Nodes represent different types of first-aid supplies, and edges represent the relationships between supplies. The knowledge graph not only contains the basic attributes of the supplies but also covers features such as the effectiveness and portability of the supplies. It is used to quickly locate first-aid supplies related to specific emergency situations. Deep learning algorithms are a type of machine learning method that can automatically learn patterns from large amounts of data and perform complex data analysis. In this solution, deep learning algorithms are used to evaluate the effectiveness and portability of each first-aid supply node relative to the user's needs in the current emergency situation. Effectiveness refers to the actual effect of the supply in dealing with the emergency situation, such as the efficiency of an ice pack in cooling or the effect of a bandage in stopping bleeding. Portability refers to whether the supply is easy to carry and use, especially in an emergency situation, which is convenient for users to quickly obtain and apply. User-specific needs include the user's special medical conditions, personal preferences, and specific limitations of the current environment.

[0144] First, according to the determined type of emergency situation, extract the first-aid supply nodes related to the emergency situation from the first-aid supply knowledge graph constructed based on the graph neural network. These nodes contain information about all supplies that may be applicable to the current emergency situation.

[0145] Second, use deep learning algorithms to analyze and process the extracted first-aid supply nodes to evaluate the evaluation results of the effectiveness and portability of each supply node relative to the user's needs in the current emergency situation. This process takes into account the applicability and practicality of the supplies in a specific situation.

[0146] Next, based on the above evaluation results and combined with the user's specific needs, perform a priority sorting process on the first-aid supply nodes to generate a sorted list of first-aid supplies. During the sorting process, the system will comprehensively consider the effectiveness, portability, and the user's needs of the supplies.

[0147] Finally, according to the sorted list of first-aid supplies, generate a list of first-aid supplies that meet the current emergency situation and the user's specific needs. This list not only ensures the effectiveness and portability of the supplies but also fully considers the user's personalized needs.

[0148] In the embodiment of this application, in an urban first-aid scenario, assume that a user feels unwell in a high-temperature environment and triggers an emergency call, and the system has determined that the type of emergency situation is "heat stroke". The following is the specific process of how the system recommends first-aid supplies:

[0149] First, the system extracts the first-aid supply nodes related to "heat stroke" from the first-aid supply knowledge graph constructed based on graph neural networks. These nodes may include ice packs, essential balm, electrolyte drinks, sun hats, etc. The knowledge graph not only provides the basic information of these supplies but also shows the relationships between them (for example, ice packs and sun hats can be used as a combination for cooling).

[0150] Secondly, the system uses deep learning algorithms to analyze and process the extracted first-aid supply nodes, evaluating the effectiveness and portability of each supply node relative to the user's needs in the current heat stroke emergency. For example, ice packs are evaluated as very effective cooling tools, while essential balm is favored for its portability. The system also takes into account the user's special needs, such as whether the user has a history of drug allergies, to avoid recommending supplies that may cause adverse reactions.

[0151] Then, based on the evaluation results and combined with the user's specific needs (such as the user has no history of allergies but has a history of hypertension), the system ranks the first-aid supply nodes in order of priority. Considering that cooling is the top priority, ice packs are listed as the highest priority; essential balm also receives a high score for its portability and immediate effect; while electrolyte drinks are ranked third because they can replenish the lost water and electrolytes. The system also excludes items containing alcohol components to avoid affecting the user's health.

[0152] Finally, based on the sorted list of first-aid supplies, the system generates a list of first-aid supplies that meets the current emergency and the user's specific needs. This list includes ice packs, essential balm, and electrolyte drinks, ensuring that users can quickly obtain appropriate first-aid support in a high-temperature environment.

[0153] Through this series of steps, the system not only provides users with a comprehensive and efficient first-aid supply recommendation but also fully considers the user's personalized needs, improving the success rate of rescue and the user experience. This method based on graph neural networks and deep learning makes the first-aid supply recommendation more intelligent and personalized, adapting to the needs of complex and changing first-aid scenarios.

[0154] This application considers that in the fields of first aid and health management, it is crucial to quickly and accurately recommend first-aid supplies suitable for the user's current emergency. Traditional first-aid supply recommendation systems often rely on preset lists or simple rule sets and are difficult to adapt to complex and changing actual environments and individual needs. To improve the accuracy and personalization of recommendations, researchers introduced a first-aid supply knowledge graph constructed based on graph neural networks and developed an intelligent recommendation algorithm that comprehensively considers the effectiveness, portability of supplies, and the user's specific needs.

[0155] This set of formulas aims to evaluate the effectiveness, portability, and match with user needs of each first aid supply node through deep learning algorithms, generating a comprehensive priority score to provide users with the most suitable first aid supply recommendations. This method not only improves first aid efficiency but also ensures the scientificity and personalization of supply selection. Therefore, a new alternative is proposed, which includes:

[0156] Optionally, according to the type of emergency situation, combining with a first aid supply knowledge graph constructed based on a graph neural network, using a deep learning algorithm to perform intelligent recommendation processing on the first aid supply information in the knowledge graph, generating a corresponding list of first aid supplies, where the recommendation process comprehensively considers the effectiveness, portability, and specific user needs of the supplies, including:

[0157] Extract first aid supply nodes related to the emergency situation from the first aid supply knowledge graph constructed based on the graph neural network according to the type of emergency situation;

[0158] Using a deep learning algorithm, analyze and process the first aid supply nodes to obtain the effectiveness E of each supply node relative to user needs in the current emergency situation i and portability P i of the evaluation results;

[0159] Through the following formula, calculate the effectiveness E of user needs i :

[0160]

[0161] where E i is the effectiveness score of the i-th first aid supply node; G i is the feature vector of the i-th first aid supply node; W eff is the weight matrix for effectiveness evaluation; b eff is the bias term for effectiveness evaluation; α 1 is the parameter controlling the slope of the effectiveness evaluation curve;

[0162] Through the following formula, calculate the portability P of user needs i :

[0163]

[0164] where P i is the portability score of the i-th first aid supply node; W port is the weight matrix for portability evaluation, used to adjust the influence of the feature vector; b port is the bias term for portability evaluation; α 2 is the parameter controlling the slope of the portability evaluation curve;

[0165] Using the user-specific demand feature vector U, calculate the matching degree D between each first-aid supply node and the user-specific demand i ;

[0166] Calculate the matching degree D between each first-aid supply node and the user-specific demand through the following formula i :

[0167]

[0168] where D i is the matching degree score between the i-th first-aid supply node and the user-specific demand; U is the feature vector of the user-specific demand; α 3 is the parameter controlling the slope of the user demand matching degree evaluation curve; b user is the bias term in the user demand matching degree evaluation;

[0169] Based on the evaluation results E i , P i and D i , use the comprehensive priority score S i to perform priority sorting on the first-aid supply nodes and generate a sorted list of first-aid supplies L;

[0170] Calculate the comprehensive priority score S through the following formula i :

[0171]

[0172] where S i is the comprehensive priority score of the i-th first-aid supply node; w 1 is the weight coefficient of the effectiveness score E i ; w 2 is the weight coefficient of the portability score P i ; w 3 is the weight coefficient of the user demand matching degree score D i ; β is the exponential parameter of the effectiveness score E i ; γ is the exponential parameter of the portability score P i ; δ is the exponential parameter of the user demand matching degree score D i ; b is the bias term for comprehensive priority calculation;

[0173] Generate the first-aid supply list L through sorting using the following formula

[0174] L = sort(S 1 , S 2 ,..., S n )

[0175] Among them, L is the sorted list of first-aid supplies; n is the total number of first-aid supply nodes; sort is the sorting function, which sorts according to the comprehensive priority score S i in descending order of value;

[0176] Generate a list of first-aid supplies that meet the current emergency situation and the specific needs of the user according to the sorted list of first-aid supplies L.

[0177] The following is a detailed annotation description of each parameter of this formula:

[0178] In the formula , S i is the comprehensive priority score of the i-th first-aid supply node, indicating the overall importance of the supply in the current emergency situation. w 1 , w 2 , w 3 are weight coefficients, representing the relative importance of the effectiveness score E i , portability score P i and the matching degree score D i of the user's needs in the comprehensive score. E i is the effectiveness score of the i-th first-aid supply node, indicating the actual effect of the supply in dealing with the emergency situation. P i is the portability score of the i-th first-aid supply node, indicating whether the supply is easy to carry and use. D i is the matching degree score of the i-th first-aid supply node with the specific needs of the user, indicating whether the supply meets the personalized needs of the user. β, γ, δ are exponential parameters, which are used to adjust the effectiveness score E i , portability score P i and the matching degree score D i of the user's needs. b is the bias term, which is used to adjust the baseline level of the comprehensive score.

[0179] The following is a brief introduction to the design reasons for each item of this formula:

[0180] In the formula for calculating the comprehensive priority score S i , This item can quantify the specific contribution of the effectiveness score E 1 to the comprehensive score by multiplying the weight coefficient w i and applying the exponential parameter β. The weight w 1 ensures that the effectiveness score accounts for an appropriate proportion in the total score, while the exponential parameter β allows further adjustment of its influence degree. This item can quantify the specific contribution of the portability score P 2 to the comprehensive score by multiplying the weight coefficient w i and applying the exponential parameter γ. The weight w 2Ensure that the portability score occupies an appropriate proportion in the total score, and the exponential parameter γ allows further adjustment of its influence degree. This sub-item is multiplied by the weight coefficient w 3 and the exponential parameter δ is applied to quantify the user demand matching degree score D i 's specific contribution to the comprehensive score. The weight w 3 Ensures that the user demand matching degree score occupies an appropriate proportion in the total score, and the exponential parameter δ allows further adjustment of its influence degree. The bias term b provides a basic comprehensive score value, so that the system can still give a reasonable default score even without the influence of other input features. This helps to avoid the problem of too low scores in extreme cases.

[0181] The following are the reasons for adding up each sub-item:

[0182] Adding up each scoring item and the bias term b is to obtain a comprehensive priority score S i . The addition operation enables the influence of each factor to be accumulated together to form an overall score. The addition of the bias term b ensures the baseline level of the score and prevents the score from being too low or too high.

[0183] Formula By comprehensively considering three key factors: effectiveness, portability, and user demand matching degree, and introducing weight coefficients and exponential parameters for flexible adjustment, the comprehensive priority score of each first aid supply node is generated. This method not only improves the accuracy and scientific nature of the recommendation, but also ensures the personalization and practicality of the material selection. This design is particularly suitable for first aid scenarios that require rapid assessment of health risks and taking corresponding measures, significantly improving the practicality and reliability of the system.

[0184] Suppose a user feels unwell in a high-temperature environment and triggers an emergency call. The system needs to recommend the most suitable first aid supplies according to the user's current situation and health condition. The following are the specific steps and calculation processes:

[0185] The following is data collection and preparation:

[0186] Type of emergency: Determined as "heat stroke".

[0187] User-specific demand feature vector U: The user has no history of allergies but has a history of hypertension.

[0188] First aid supply node feature vector G i : Relevant features (such as cooling effect, portability, etc.) of supplies such as ice packs, essential balm, and electrolyte drinks.

[0189] Calculate the effectiveness score E i: For the ice pack (the first supply node), assume its feature vector G 1 = [0.8, 0.7], where the first value represents the cooling effect and the second value represents other characteristics (such as material quality). Calculate using the following parameters:

[0190] Weight matrix W eff = [0.6, 0.4]; Bias term b eff = 0.1; Slope parameter α 1 = 2;

[0191]

[0192] Calculate the user demand matching degree D i : For the ice pack (the first supply node), assume the user's specific demand feature vector U = [0.9, 0.8], and calculate using the following parameters:

[0193] Slope parameter α 3 = 1; Bias term b user = 0.1;

[0194]

[0195] Calculate the comprehensive priority score S i : Assume for the ice pack (the first supply node), calculate the comprehensive priority score using the following parameters:

[0196] Weight coefficient w 1 = 0.5, w 2 = 0.3, w 3 = 0.2; Exponential parameters β = 1, γ = 1, δ = 1; Bias term b = 0.1;

[0197]

[0198] S 1 = 0.5·0.70 1 + 0.3·0.72 1 + 0.2·0.79 1 + 0.1 ≈ 0.824

[0199] Sort to generate the first-aid supply list L: Assume there are three first-aid supply nodes (ice pack, essential balm, electrolyte drink), calculate their comprehensive priority scores S i , and sort to generate the final first-aid supply list L:

[0200] Ice pack S 1 ≈ 0.824; Essential balm S 2 ≈ 0.78; Electrolyte drink S 3 ≈ 0.75;

[0201] L = sort(S 1 , S 2 , S 3 ); L = [ice pack, essential balm, electrolyte drink]

[0202] According to the above calculation, the first-aid supply with the highest comprehensive priority score is the ice pack, followed by the essential balm, and finally the electrolyte drink.

[0203] Immediate response: The ice pack is listed as the highest priority because of its high effectiveness and portability, and it meets the needs of users. It can quickly reduce body temperature and relieve the symptoms of heatstroke.

[0204] Auxiliary measure: The essential balm ranks second due to its portability and immediate effect. It can provide an immediate sense of coolness and help users relieve discomfort.

[0205] Long-term support: Although the electrolyte drink is important, it is not as urgent as the previous two, so it ranks third. It can supplement the lost water and electrolytes and help restore physical strength.

[0206] Through this intelligent recommendation method based on graph neural network and deep learning, the system not only provides users with a comprehensive and efficient first-aid supply recommendation, but also fully considers the personalized needs of users, improving the success rate of rescue and user experience. This method is particularly suitable for complex traffic and changing first-aid needs in urban environments, providing users with more personalized and efficient first-aid solutions.

[0207] 104. Sort the first-aid supply list according to the importance and availability of items to obtain a first-aid supply list arranged by priority, and display it to the user through the user interface using augmented reality technology. At the same time, adjust the navigation information of the nearest first-aid supply storage location based on real-time traffic data.

[0208] In this step, the generated first-aid supply list is sorted by priority, mainly based on the importance and availability of supplies. Importance reflects the key role of supplies in dealing with emergencies, while availability refers to the ease of obtaining supplies in the current environment. The sorted list is intuitively displayed to the user through augmented reality technology to help the user quickly understand the recommended content. In addition, the system will also adjust the navigation information of the nearest first-aid supply storage location based on real-time traffic data to ensure that the user can find the required supplies most quickly.

[0209] The system prioritizes the first-aid supply list according to the importance and availability of supplies. For each supply, the system evaluates its crucial role in dealing with emergencies and checks the inventory at nearby first-aid supply storage points. After the prioritization is completed, the system uses augmented reality technology to visually display the recommended supply list to the user, enabling the user to directly view the recommended supplies and their priorities on the mobile phone screen. Meanwhile, based on real-time traffic data, the system dynamically adjusts the navigation route to guide the user to the nearest and most convenient first-aid supply storage point.

[0210] In the example of this application, after generating a first-aid supply list including ice packs, essential balm, and electrolyte drinks, the system prioritizes it according to the importance and availability of supplies. Considering that cooling down is the top priority, the ice packs are listed as the highest priority. The system checks the inventory at nearby pharmacies and first-aid stations and finds that a pharmacy has an adequate supply of ice packs. Then, the system uses augmented reality technology to display the recommended supplies and their priorities on the user's mobile phone interface and generates a navigation route. The navigation information is optimized based on real-time traffic data to guide the user to avoid congested roads and quickly reach the pharmacy to obtain the ice packs and other recommended supplies, ensuring that the user can obtain effective first-aid support in the shortest time.

[0211] Optionally, the step of prioritizing the first-aid supply list according to the importance and availability of items in step 104 to obtain a first-aid supply list arranged by priority and using augmented reality technology to display it to the user through the user interface, and at the same time adjusting the navigation information of the nearest first-aid supply storage location based on real-time traffic data includes: evaluating each supply in the list based on the importance of each item in the first-aid supply list and its availability in the current environment to generate a comprehensive priority for each supply; sorting the first-aid supply list according to the comprehensive priority of each supply to obtain a first-aid supply list arranged by priority; using augmented reality technology to visually display the first-aid supply list arranged by priority to the user through the user interface; determining the best route to the nearest first-aid supply storage point based on the user's current location, combined with the first-aid supply list arranged by priority and real-time traffic data, and providing dynamically updated navigation information.

[0212] In this step, importance refers to the key role of supplies in dealing with emergencies. For example, in the case of heat stroke, the importance of cooling tools (such as ice packs) is much higher than that of other non-essential items. Availability refers to the ease of obtaining supplies in the current environment. This includes the inventory situation at nearby storage points, whether the supplies are easy to carry, etc. Availability also takes into account traffic conditions and distance factors. The comprehensive priority combines the two dimensions of importance and availability to evaluate each item of supplies, generating a comprehensive score for ranking. Augmented reality technology superimposes virtual information onto the real-world view through cameras and sensors, enabling users to visually see the recommended first-aid supplies and their priorities. This not only improves the user experience but also guides users to find the required supplies more quickly. Real-time traffic data includes information such as road congestion conditions and public transportation schedules, which are used to dynamically adjust the navigation route to ensure that users can reach the nearest first-aid supply storage point most quickly.

[0213] First, based on the importance of each item in the first-aid supply list and its availability in the current environment, each item in the list is evaluated to generate the comprehensive priority of each supply. During the evaluation process, the system comprehensively considers factors such as the key role of the supplies, inventory situation, and transportation convenience.

[0214] Second, according to the comprehensive priority of each supply, the first-aid supply list is sorted to obtain a first-aid supply list arranged by priority. This sorting ensures that users can obtain the most important supplies first.

[0215] Next, using augmented reality technology, the first-aid supply list arranged by priority is visually displayed to the user through the user interface. Users can intuitively view the recommended supplies and their priorities on the mobile phone screen, which is convenient for quick understanding and taking action.

[0216] Finally, based on the user's current location, combined with the first-aid supply list arranged by priority and real-time traffic data, the best route to the nearest first-aid supply storage point is determined, and dynamic updated navigation information is provided. The navigation information is adjusted in real time according to traffic conditions to ensure that users can reach the target location in the shortest time.

[0217] In the embodiment of this application, in an urban first-aid scenario, assume that a user feels unwell in a high-temperature environment and triggers an emergency call. The system has generated a first-aid supply list containing ice packs, essential balm, and electrolyte drinks. The following is the specific process of how the system performs priority sorting and navigation:

[0218] First, the system evaluates the importance and availability of each item. Ice packs are evaluated as the highest priority because they have an immediate cooling effect and are in stock at nearby pharmacies; essential balm ranks second due to its portability and the effect of instantly relieving discomfort; electrolyte drinks, although important, are not as urgent as the previous two, so they rank third. The system also checks the inventory of nearby first aid stations and pharmacies to ensure that the recommended items are indeed available.

[0219] Secondly, based on the above evaluation results, the system sorts the first aid supply list and generates a first aid supply list arranged by priority. The final order is: ice pack > essential balm > electrolyte drink. This sorting ensures that users can obtain the most important items first.

[0220] Next, the system uses augmented reality technology to visually display the first aid supply list arranged by priority to the user through the user's smartphone camera. The user can directly see the recommended items and their priorities on the screen, and even see pictures and brief descriptions of the items. This intuitive display method helps users quickly understand the recommendations and take corresponding measures.

[0221] Finally, based on the user's current location (determined by GPS), the system combines real-time traffic data to determine the best route to the nearest first aid supply storage point. For example, the system finds that a certain pharmacy near the user has an adequate stock of ice packs, so it generates a navigation path to guide the user to that pharmacy. The navigation information is dynamically updated according to the real-time traffic conditions to avoid traffic jams for the user and ensure that they can obtain the required first aid supplies in the shortest possible time.

[0222] Through this series of steps, the system not only provides users with a comprehensive and efficient first aid supply recommendation, but also optimizes the supply acquisition process through augmented reality technology and real-time traffic data. This intelligent recommendation and navigation method significantly improves the first aid efficiency and user experience, ensuring that users can quickly obtain appropriate first aid support in case of emergencies. This method is particularly suitable for complex traffic and changing first aid needs in urban environments, providing users with a more personalized and efficient first aid solution.

[0223] Figure 2 A schematic structural diagram of an AI-based first aid supply recommendation system is provided for the embodiments of this application, as Figure 2 shown. The system includes:

[0224] An acquisition and integration module 21, configured to acquire real-time data of the user's current environment and the user's physiological state information, and perform integration processing on the real-time data and the user's physiological state information to obtain integrated user environment and health condition data;

[0225] An analysis and determination module 22, configured to analyze the correlation between the environmental factors around the user and the personal health status through a context-aware algorithm based on the integrated user environment and health status data, and then determine the type of emergency encountered by the user by using a pre-trained multi-modal fusion model;

[0226] A processing and generation module 23, configured to perform intelligent recommendation processing on the first-aid supply information in the knowledge graph by using a deep learning algorithm according to the type of emergency, in combination with a first-aid supply knowledge graph constructed based on a graph neural network, to generate a corresponding first-aid supply list, where the recommendation process comprehensively considers the effectiveness, portability of the supplies, and the specific needs of the user;

[0227] A sorting and navigation module 24, configured to perform priority sorting on the first-aid supply list according to the importance and availability of the items to obtain a first-aid supply list arranged by priority, and display it to the user through a user interface by using augmented reality technology, and at the same time adjust the navigation information of the nearest first-aid supply storage location based on real-time traffic data.

[0228] Figure 2 The AI-based first-aid supply recommendation system described above can execute Figure 1 The AI-based first-aid supply recommendation method described in the illustrated embodiment, and its implementation principle and technical effects will not be elaborated. For the AI-based first-aid supply recommendation system in the above embodiment, the specific manners in which each module and unit perform operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0229] In a possible design, Figure 2 The AI-based first-aid supply recommendation system in the illustrated embodiment can be implemented as a computing device, such as Figure 3 shown, and this computing device may include a storage component 31 and a processing component 32;

[0230] The storage component 31 stores one or more computer instructions, where the one or more computer instructions are called and executed by the processing component 32.

[0231] The processing component 32 is used to: obtain real-time data of the user's current environment and the user's physiological state information, integrate and process the real-time data and the user's physiological state information to obtain integrated user environment and health status data; based on the integrated user environment and health status data, analyze the correlation between the environmental factors around the user and the personal health status through a context awareness algorithm, and then use a pre-trained multi-modal fusion model to determine the type of emergency encountered by the user; according to the type of emergency, combined with the first-aid supply knowledge graph constructed based on a graph neural network, use a deep learning algorithm to perform intelligent recommendation processing on the first-aid supply information in the knowledge graph to generate a corresponding first-aid supply list, where the recommendation process comprehensively considers the effectiveness, portability of the supplies and the specific needs of the user; perform a priority sorting process on the first-aid supply list according to the importance and availability of the items to obtain a first-aid supply list arranged by priority, and use augmented reality technology to display it to the user through the user interface, and at the same time adjust the navigation information of the nearest first-aid supply storage location based on real-time traffic data.

[0232] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0233] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0234] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0235] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0236] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0237] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server. The above-mentioned processing components, storage components, etc. can be basic server resources leased or purchased from a cloud computing platform.

[0238] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above-mentioned Figure 1 AI-based first-aid supplies recommendation method shown in the embodiment.

[0239] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be described herein again.

[0240] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0241] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0242] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. An AI-based method for recommending emergency supplies, characterized in that: include: Acquire real-time data of the user's current environment and information about the user's physiological status, integrate the real-time data and the user's physiological status information, and obtain integrated user environment and health status data; Based on the integrated user environment and health status data, the correlation between the user's surrounding environmental factors and personal health status is analyzed by a context-aware algorithm, and then the type of emergency situation encountered by the user is determined by using a pre-trained multimodal fusion model; According to the type of emergency, combined with the first aid supplies knowledge graph built based on the graph neural network, a deep learning algorithm is used to perform intelligent recommendation processing on the first aid supplies information in the knowledge graph to generate a corresponding first aid supplies list, wherein the recommendation process comprehensively considers the effectiveness, portability and user-specific needs of the supplies; The first aid supplies list is prioritized according to the importance and availability of the items to obtain a prioritized first aid supplies list, which is displayed to the user through a user interface using augmented reality technology, and navigation information of the nearest first aid supplies storage location is adjusted based on real-time traffic data.

2. The method according to claim 1, characterized in that Based on the integrated user environment and health status data, the correlation between the user's surrounding environmental factors and personal health status is analyzed by a context awareness algorithm, and then the type of emergency situation encountered by the user is determined by using a pre-trained multimodal fusion model, including: Based on the integrated user environment and health data, the improved context-aware algorithm combined with the Bayesian network is used to dynamically model and comprehensively analyze the user's real-time location information, environmental factors such as ambient temperature and humidity, and physiological status information such as heart rate and blood pressure, to obtain the user's current specific situation mode and potential change trend; According to the specific situation patterns and potential change trends, they are input into the situation-health influencing factor prediction model based on transfer learning optimization to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate health risk scores; Based on the health risk score, reinforcement learning strategies are used to simulate the best response measures in different emergency situations. Combined with the frequency and severity statistics of emergencies in the historical case library, a pre-trained and adaptively adjusted multimodal fusion model is used to further refine the analysis and determine the specific types of emergencies that users may encounter and their urgency.

3. The method according to claim 2, characterized in that Based on the integrated user environment and health status data, the improved context-aware algorithm combined with the Bayesian network is used to dynamically model and comprehensively analyze the user's real-time location information, environmental factors of ambient temperature and humidity, and physiological status information of heart rate and blood pressure, to obtain the user's current specific situation mode and potential change trend, including: Based on the integrated user environment and health status data, the improved context-aware algorithm is used to identify key environmental factors and physiological indicators; Based on the key environmental factors and physiological indicators, a probabilistic reasoning process is performed through a constructed Bayesian network model to quantify the interactions between these factors and their impact on the user's health status; Based on the probability distribution output by the Bayesian network model, combined with the time series analysis method, the changing trend of the user's environment and health status in the future is predicted and processed to generate a change trend prediction result; By utilizing the change trend prediction result and performing pattern recognition processing on the probability distribution and the change trend, the specific situation mode in which the user is currently located is determined, and the future development direction of the specific situation mode is generated.

4. The method according to claim 2, characterized in that: The specific situation patterns and potential change trends are input into the situation-health impact factor prediction model based on transfer learning optimization to evaluate the immediate and long-term health risks caused by different combinations of environmental factors and physiological states, and generate health risk scores, including: According to the specific situation patterns and potential change trends, a data set containing the user's real-time location information, environmental factors of ambient temperature and humidity, and physiological status information of heart rate and blood pressure is constructed; Based on the constructed data set, it is input into the context-health impact factor prediction model based on transfer learning optimization for processing to obtain preliminary health impact analysis results; Using the preliminary health impact analysis results, further analyze and process the immediate and long-term health impacts under different combinations of environmental factors and physiological states to generate a detailed health impact assessment; Based on the detailed health impact assessment, a health risk score corresponding to each combination of environmental factors and physiological states is calculated and generated.

5. The method according to claim 2, characterized in that: Based on the health risk score, the reinforcement learning strategy is used to simulate the best response measures in different emergency situations. Combined with the statistics of emergency frequency and severity in the historical case library, the pre-trained and adaptively adjusted multimodal fusion model is used to further refine the analysis and determine the specific emergency types and urgency that users may encounter, including: Based on the health risk score, the reinforcement learning algorithm is used to simulate various response measures taken in different emergency situations to obtain the best response set for each emergency situation; According to the optimal response measure set, combined with the emergency frequency and severity statistics recorded in the historical case library, the optimal response measure set is weighted and adjusted to obtain an adjusted optimal response measure set; Based on the adjusted optimal response set and relevant historical data, further analysis and processing are performed using a pre-trained and adaptively adjusted multimodal fusion model to generate a prediction result; Based on the prediction results, the specific type of emergency situation encountered by the user is determined, and the urgency of each emergency situation is determined by evaluating the probability of occurrence and potential impact of each emergency situation.

6. The method according to claim 1, characterized in that According to the type of emergency, combined with the first aid supplies knowledge graph built based on the graph neural network, a deep learning algorithm is used to perform intelligent recommendation processing on the first aid supplies information in the knowledge graph to generate a corresponding first aid supplies list, wherein the recommendation process comprehensively considers the effectiveness, portability and user-specific needs of the supplies, including: According to the type of emergency, extract the first aid material nodes related to the emergency from the first aid material knowledge graph built based on the graph neural network; Analyze and process the emergency material nodes using a deep learning algorithm to obtain an evaluation result of the effectiveness and portability of each material node relative to user needs in the current emergency situation; Based on the evaluation results and in combination with the specific needs of the user, the first aid material nodes are prioritized and a sorted first aid material list is generated; According to the sorted first aid material list, a first aid material list that meets the current emergency situation and conforms to the user's specific needs is generated.

7. The method according to claim 1, characterized in that The first aid supplies list is prioritized according to the importance and availability of the items to obtain a prioritized first aid supplies list, and is displayed to the user through a user interface using augmented reality technology, while adjusting the nearest first aid supplies storage location navigation information based on real-time traffic data, including: Based on the importance of each item in the emergency supplies list and its availability in the current environment, each item in the list is evaluated and processed to generate a comprehensive priority for each item; Sorting the list of first aid supplies according to the comprehensive priority of each supply to obtain a list of first aid supplies arranged by priority; Using augmented reality technology, the priority-arranged list of emergency supplies is presented to the user in a visual form through a user interface; Based on the user's current location, combined with the prioritized list of emergency supplies and real-time traffic data, the best route to the nearest emergency supply storage point is determined, and dynamically updated navigation information is provided.

8. An AI-based emergency supplies recommendation system, characterized in that: include: An acquisition and integration module is used to acquire real-time data of the user's current environment and information on the user's physiological status, and integrate the real-time data and the user's physiological status information to obtain integrated user environment and health status data; An analysis and determination module is used to analyze the correlation between the user's surrounding environmental factors and personal health status through a context-aware algorithm based on the integrated user environment and health status data, and then use a pre-trained multimodal fusion model to determine the type of emergency situation encountered by the user; A processing and generating module, which is used to perform intelligent recommendation processing on the first aid material information in the knowledge graph based on the emergency type and the first aid material knowledge graph constructed based on the graph neural network, and generate a corresponding first aid material list, wherein the recommendation process comprehensively considers the effectiveness, portability and user-specific needs of the materials; The sorting and navigation module is used to prioritize the first aid supplies list according to the importance and availability of the items, obtain a prioritized first aid supplies list, and display it to the user through the user interface using augmented reality technology, while adjusting the navigation information of the nearest first aid supplies storage location based on real-time traffic data.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an AI-based first aid material recommendation method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, an AI-based first aid material recommendation method as described in any one of claims 1 to 7 is implemented.

Images