Satellite-borne AI emergency medical inquiry method and system

Abstract

The application provides a kind of on-orbit AI emergency medical inquiry method and system, applied to satellite.Receiving user terminal oral medical inquiry short message, analysis to extract symptom characteristics, user location information and environmental information, generate structured inquiry data adapted to clinical triage logic.Call on-orbit lightweight medical inquiry AI model decision to generate medical guidance text, then encode the text into reply short message and issue to user terminal.The application preposes oral understanding, clinical structuring and medical reasoning whole process to satellite end, without ground station and data center, compresses response time to meet emergency golden window.On-orbit AI model can accurately understand oral injury, automatically map clinical triage logic, and handle symptom diversity and drug contraindications and other professional barrier problems, avoid false self-help.The whole inquiry closed loop does not depend on ground medical resources and communication network, can provide instant, professional and reliable emergency medical support for network-free areas.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a spaceborne AI emergency medical consultation method and system. Background Technology

[0002] With the rapid development of satellite communication technology, global emergency communication and rescue capabilities have been significantly improved. In existing satellite emergency communication systems, IoT satellites and similar systems generally employ a store-and-forward mode. Users send distress messages via handheld terminals or wearable devices. The satellite receives and buffers the data as it passes overhead, then transmits it to the ground control center after passing through the ground station's coverage area. The ground system then manually interprets the original messages or processes them using cloud algorithms before transmitting basic rescue instructions or location confirmation information back to the user terminal via the satellite link.

[0003] However, the existing solutions mentioned above have the following drawbacks in emergency medical care: First, there is a high delay in medical consultation response. The golden rescue time for emergencies such as cardiac arrest, severe trauma, and altitude sickness is usually only a few minutes to tens of minutes, while consultation requests need to go through multiple stages, including satellite overpass ground stations, ground data center processing, and feedback transmission, which often takes tens of minutes to several hours. This delay can directly cause patients to miss the best treatment window, endangering their lives. Second, there is a lack of professional medical guidance output. Medical consultation involves highly specialized knowledge such as the diversity of symptom descriptions, disease differential diagnosis, drug contraindications, and environmental adaptation management. However, the existing system can only forward coordinates and simple text, and does not have any medical semantic understanding and reasoning capabilities. It cannot provide targeted treatment references for trapped personnel. Trapped personnel may aggravate their injuries or even cause irreversible consequences by making incorrect self-rescue attempts while waiting. Third, it is highly dependent on ground medical resources and networks. After the information is transmitted to the ground station, it needs to be analyzed by qualified medical personnel before being transmitted back. Once ground communication is interrupted, medical resources are overwhelmed, or the medical system in the disaster area is paralyzed, the entire rescue chain will be completely cut off.

[0004] It is evident that in emergency medical scenarios where there is no terrestrial network coverage or terrestrial facilities are damaged, the existing satellite communication architecture is insufficient to meet the actual needs of emergency medical consultations. How to break through the technical bottlenecks of the traditional store-and-forward mode and improve the response time and professional medical guidance capabilities of emergency medical consultations has become an urgent technical problem to be solved in this field. Summary of the Invention

[0005] This invention provides a spaceborne AI emergency medical consultation method and system to overcome the technical defects of existing satellite emergency communication systems in emergency medical care, such as high response delay, lack of professional medical guidance, and high dependence on ground medical resources and networks.

[0006] This invention provides a spaceborne AI-based emergency medical consultation method, applied to satellites, comprising: The system receives short medical consultation messages containing colloquial descriptions of injuries sent by user terminals, parses the short medical consultation messages, extracts symptom features, user location information, and environmental information, and generates structured consultation data that adapts to clinical triage logic. The lightweight medical consultation AI model deployed on the satellite is invoked to perform decision-making reasoning based on the structured consultation data and generate medical guidance text. The medical guidance text is encoded into a short reply message and sent to the user terminal via a downlink.

[0007] According to the present invention, a spaceborne AI emergency medical consultation method is provided, wherein parsing the medical consultation short message, extracting symptom features, user location information, and environmental information, and generating structured consultation data adapted to clinical triage logic includes: The medical consultation short message is cleaned and semantically segmented to identify the symptom features, user location information, and environmental information. After the identified information is normalized and formatted, it is mapped to predefined standardized fields to generate the structured consultation data adapted to the clinical triage logic.

[0008] According to a satellite-borne AI emergency medical consultation method provided by the present invention, the step of calling a lightweight medical consultation AI model deployed on the satellite, performing decision reasoning based on the structured consultation data, and generating medical guidance text includes calling the lightweight medical consultation AI model to perform the following steps: The symptom features in the structured consultation data are semantically vectorized to generate symptom semantic vectors; Using the symptom semantic vector as the retrieval criteria, similarity matching is performed in the spaceborne emergency medical knowledge base to retrieve the associated clinical decision-making paths and drug contraindications. The clinical decision-making path, the drug contraindications, the user location information, and the environmental information in the structured consultation data are concatenated to construct a multi-source reasoning context; Based on the multi-source reasoning context, multi-source information fusion reasoning is performed to generate the medical guidance text, which includes personalized triage suggestions, environment-adaptive emergency operation steps, and medication contraindication reminders.

[0009] According to a satellite-borne AI emergency medical consultation method provided by the present invention, before invoking the lightweight medical consultation AI model deployed on the satellite, the method further includes: The structured consultation data is matched with keywords indicating urgency. If a preset high-risk emergency keyword is matched, a high-priority marker is generated for the current medical consultation short message. Based on the high-priority label, the current structured consultation data is prioritized for scheduling, enabling the lightweight medical consultation AI model to prioritize decision-making and reasoning based on the current structured consultation data.

[0010] The spaceborne AI emergency medical consultation method provided by the present invention further includes: When a subsequent medical consultation short message is received from the user terminal, the consultation dialogue history of the same user terminal stored by the satellite during multiple transit cycles is obtained. The subsequent medical consultation short messages are parsed to generate subsequent structured consultation data; The subsequent structured consultation data is concatenated with the consultation dialogue history to construct a multi-round consultation context; The lightweight medical consultation AI model is invoked to perform decision-making reasoning based on the multi-round consultation context and generate continuous medical guidance text. The continuous medical guidance text is encoded into the reply short message and sent to the user terminal via the downlink to respond to the subsequent medical consultation short message.

[0011] The spaceborne AI emergency medical consultation method provided by the present invention further includes: The medical consultation short message, the medical guidance text, and the processing timestamp are packaged into consultation record data and written into the onboard non-volatile memory. When the satellite flies over the communication coverage area of ​​the ground station, it will download the consultation record data in batches to the ground rescue system for medical dispatch and medical record archiving.

[0012] According to the present invention, a spaceborne AI emergency medical consultation method is provided, wherein the lightweight medical consultation AI model is trained based on large-scale clinical dialogue data, emergency guidelines and drug contraindication knowledge base, and is adapted to the computing power and storage constraints of the satellite embedded processor by model pruning, quantization or knowledge distillation compression.

[0013] This invention also provides a spaceborne AI emergency medical consultation device, applied to a satellite, comprising: The data receiving and parsing module is used to receive medical consultation short messages containing colloquial descriptions of injuries sent by user terminals, and to parse the medical consultation short messages to extract symptom features, user location information and environmental information, and generate structured consultation data that is adapted to clinical triage logic. The decision reasoning module is used to call the lightweight medical consultation AI model deployed on the satellite, perform decision reasoning based on the structured consultation data, and generate medical guidance text; The text delivery module is used to encode the medical guidance text into a short reply message and send it to the user terminal via a downlink.

[0014] The present invention also provides an Internet of Things satellite, comprising: an onboard memory, an onboard processor, and a computer program stored on the onboard memory and running on the onboard processor, wherein the onboard processor, when executing the computer program, implements any of the above-mentioned onboard AI emergency medical consultation methods.

[0015] The present invention also provides a spaceborne AI emergency medical consultation system, including: a user terminal and the aforementioned Internet of Things satellite; The user terminal is connected to the IoT satellite via a satellite-to-ground link for sending short messages for medical consultations and receiving short reply messages from the IoT satellite.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the spaceborne AI emergency medical consultation method as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the spaceborne AI emergency medical consultation method as described above.

[0018] This invention provides a satellite-based AI-powered emergency medical consultation method and system. The method, applied to a satellite, first receives short medical consultation messages containing colloquial descriptions of injuries from user terminals. It then analyzes and extracts symptom features, location information, and environmental information to generate structured consultation data adapted to clinical triage logic. Next, it invokes a lightweight medical consultation AI model deployed on the satellite to perform decision-making reasoning and generate medical guidance text. Finally, the text is encoded into a reply message and sent to the user terminal via downlink. This invention executes the entire process of colloquial understanding, clinical structured conversion, and medical reasoning on the satellite, eliminating the need for ground stations and data centers. This reduces consultation response time from tens of minutes to several hours to within the satellite's transit cycle, meeting the critical window for emergency rescue. Furthermore, the onboard AI model achieves accurate understanding of colloquial injuries and automatic mapping to clinical triage logic, handling highly specialized issues such as symptom diversity and drug contraindications. It effectively addresses the fundamental deficiency of existing systems that can only forward coordinates and simple text, lacking medical semantic understanding capabilities, thus preventing trapped individuals from aggravating their injuries due to incorrect self-rescue attempts. Meanwhile, the entire consultation interaction is completed in a closed loop on the satellite, without relying on ground medical resources and communication networks. It can still operate autonomously in extreme scenarios such as damage to ground facilities or communication interruption, providing immediate, professional and reliable emergency medical support for areas without ground network coverage. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts of the spaceborne AI emergency medical consultation method provided by the present invention.

[0021] Figure 2 This is the second flowchart illustrating the spaceborne AI emergency medical consultation method provided by the present invention.

[0022] Figure 3 This is the third flowchart illustrating the spaceborne AI emergency medical consultation method provided by the present invention.

[0023] Figure 4 This is a timing comparison diagram of the spaceborne AI emergency medical consultation method provided by the present invention and the traditional store-and-forward mode.

[0024] Figure 5 This is a schematic diagram of the structure of the spaceborne AI emergency medical consultation device provided by the present invention.

[0025] Figure 6 This is a schematic diagram of the structure of the Internet of Things satellite provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] Existing satellite emergency communication systems generally employ a store-and-forward model. Medical consultation messages sent by users must pass through satellite overpass ground stations, be processed by ground data centers, and analyzed by medical personnel before being transmitted back. This results in response delays of tens of minutes to several hours, failing to meet the critical rescue window of just a few minutes required in emergencies such as cardiac arrest and severe injuries. Furthermore, existing systems can only forward coordinates and simple text, unable to understand colloquial descriptions of injuries, and lack the medical expertise to identify symptoms, make differential diagnoses, or determine drug contraindications. Trapped individuals often worsen their injuries due to incorrect self-rescue methods. In addition, the entire rescue chain is highly dependent on ground medical resources and networks; once ground communication is interrupted or the medical system collapses, consultation and interaction will be completely disrupted.

[0028] To address the aforementioned problems in existing technologies, this invention provides a spaceborne AI-powered emergency medical consultation method and system. The inventive concept involves deploying a lightweight medical consultation AI model on a satellite, utilizing the satellite's onboard processing capabilities to form a closed-loop medical consultation interaction link between the user terminal and the satellite. Specifically, after receiving a short medical consultation message containing a colloquial description of the injury from the user terminal, the satellite completes the entire process onboard, including natural language parsing, clinical triage logic mapping, structured consultation data generation, AI medical reasoning, and the encoding and distribution of the reply message. This process eliminates the need for ground stations, data centers, or ground medical personnel, enabling real-time onboard response and professional guidance for emergency medical consultations. This fundamentally solves the problems of long delays, lack of professionalism, and network dependence inherent in traditional solutions that rely on ground facilities.

[0029] The following detailed description of the spaceborne AI emergency medical consultation method and system provided by the present invention, in conjunction with the accompanying drawings and specific embodiments, is provided in detail.

[0030] Figure 1 This is one of the flowcharts illustrating the spaceborne AI emergency medical consultation method provided by the present invention. This method is applied to satellites, such as... Figure 1 As shown, the spaceborne AI emergency medical consultation method provided by this invention includes: S101. Receive a short medical consultation message containing a colloquial description of the injury sent by the user terminal, parse the short medical consultation message, extract symptom features, user location information and environmental information, and generate structured consultation data adapted to clinical triage logic.

[0031] The user terminal can be a short message transceiver module integrated into a Beidou handheld device, satellite phone, emergency distress beacon, or smart wearable device. It has text input and output functions and supports users to send medical consultation short messages in natural language.

[0032] Specifically, the satellite receives medical consultation short messages sent by user terminals via an uplink communication link. Users can describe their injuries in everyday language, such as "Right shinbone fracture, bleeding uncontrollably, I have a triangular bandage, how do I stop the bleeding?", or "My stomach hurts a lot, I've vomited twice, I have no medicine," etc. After receiving the medical consultation short messages containing colloquial injury descriptions from user terminals, the satellite performs semantic analysis on the messages, extracting symptom features such as fracture, user location information, and environmental information. User location information can be extracted from the medical consultation short messages or carried by the user terminal itself. Environmental information includes descriptions of environmental parameters such as being in the wild, high temperature, or heavy rain. Further, the extracted information is processed according to a predefined clinical triage logic template to generate structured consultation data objects adapted to clinical triage logic, which are called structured consultation data. This structured data facilitates efficient processing by subsequent AI models and conforms to clinical decision-making logic.

[0033] In some embodiments, step S101, which involves parsing the medical consultation short message, extracting symptom features, user location information, and environmental information, and generating structured consultation data adapted to clinical triage logic, may include: The system performs text cleaning and semantic segmentation on short medical consultation messages, identifies symptom features, user location information, and environmental information, and then normalizes and aligns the identified information to predefined standardized fields to generate structured consultation data that is adapted to clinical triage logic.

[0034] First, the medical consultation short messages undergo text cleaning and semantic segmentation. Text cleaning aims to remove irrelevant symbols, stop words, and repetitive phrases. Semantic segmentation divides the cleaned natural language sentences into independent semantic units. For example, "right calf fracture, persistent bleeding, triangular bandage" is segmented into three semantic units: "right calf fracture," "persistent bleeding," and "triangular bandage." Then, using a pre-defined named entity recognition model, entity information such as symptom features, user location information, and environmental information is identified from each semantic unit.

[0035] Secondly, the identified information is normalized and formatted. Entity normalization refers to mapping different expressions of the same concept to standard medical terminology. For example, "leg," "leg part," and "lower limb" are unified as "lower limb," and "uncontrollable bleeding" and "massive bleeding" are unified as "active massive bleeding." Formatting alignment refers to standardizing the extracted values ​​and units. For example, time descriptions are unified as minutes, and temperature is unified as degrees Celsius.

[0036] Finally, the normalized and aligned information is mapped to predefined standardized fields. These standardized fields are designed according to clinical triage logic; for example, "state of consciousness," "respiratory status," and "circulatory signs" are set as the highest priority fields, while "location of injury" and "pain level" are set as secondary fields. Environmental factors and other auxiliary decision-making fields are also retained. This mapping generates a structured medical data object with complete fields and unique values, thus achieving a complete adaptation to clinical triage logic at the data level and providing medically logical input for subsequent AI model decision-making and reasoning.

[0037] S102. Invoke the lightweight medical consultation AI model deployed on the satellite, perform decision-making reasoning based on structured consultation data, and generate medical guidance text.

[0038] After generating structured consultation data, the onboard lightweight medical consultation AI model is activated for decision-making and reasoning. Specifically, the lightweight medical consultation AI model takes structured consultation data as input. For example, it can use the logic of clinical decision trees or medical reasoning engines to comprehensively analyze symptom characteristics, user location information, and environmental information to determine the possible types of diseases or injuries, assess the degree of urgency, and retrieve corresponding first aid measures, medication suggestions, and self-help contraindications, ultimately generating a medical guidance text for the user.

[0039] The lightweight medical consultation AI model can be pre-deployed on the satellite's onboard embedded processor and trained based on large-scale clinical dialogue data, first aid guidelines, and drug contraindication knowledge bases, enabling it to possess medical capabilities such as symptom understanding, injury reasoning, and first aid guidance. Furthermore, to ensure the lightweight medical consultation AI model can adapt to the limited computing and storage resources of the satellite payload, model compression techniques such as pruning, quantization, or knowledge distillation can be used to lightweight it before deployment, allowing it to run in real-time on the satellite. This invention does not limit the specific implementation methods of the training of the lightweight medical consultation AI model or the compression techniques used.

[0040] It should also be noted that the satellite involved in this invention refers to an on-orbit spacecraft with short message communication and on-board processing capabilities. It can receive medical consultation short messages sent by user terminals and independently complete parsing, reasoning, and response delivery on-board, without relying on ground stations or data centers. Satellite types may include, but are not limited to, IoT satellites, low-Earth orbit communication satellites, and navigation satellites, and can be flexibly selected according to actual deployment conditions.

[0041] In some embodiments, step S102 may be implemented by calling a lightweight medical consultation AI model to perform, such as Figure 2 The steps shown are as follows: Figure 2This is the second flowchart illustrating the spaceborne AI emergency medical consultation method provided by the present invention, as shown below. Figure 2 As shown, it includes: S1021. Perform semantic vectorization representation on the symptom features in the structured consultation data to generate symptom semantic vectors.

[0042] The lightweight medical consultation AI model extracts symptom feature fields from structured consultation data, such as fractures and uncontrollable bleeding, and transforms these discrete symptom descriptions into numerical vectors in a high-dimensional continuous space, namely symptom semantic vectors, so as to reflect the semantic relationship between symptoms and facilitate subsequent similarity calculation with symptom templates in the knowledge base.

[0043] S1022. Using symptom semantic vectors as search criteria, perform similarity matching in the spaceborne emergency medical knowledge base to retrieve associated clinical decision-making paths and drug contraindications.

[0044] The onboard emergency medical knowledge base pre-stores emergency procedures, recommended medications, contraindications, and environmental adaptation rules for common injuries and illnesses. Each knowledge entry can be indexed using semantic vectors. The lightweight medical consultation AI model uses symptom semantic vectors as query conditions to search the knowledge base for entries most similar to the current symptoms, filtering out the most relevant clinical decision-making paths and corresponding medication contraindications. This retrieval process is completed in real time on the satellite, without relying on ground-based databases.

[0045] It should be noted that a clinical decision-making pathway refers to standardized diagnostic and treatment procedures for specific injuries or symptoms, such as a sequential decision-making logic from symptom identification, emergency treatment, medication guidance to follow-up observation. Drug contraindications refer to a list of drugs that are prohibited or require cautious use under specific injuries or symptoms, along with related conditions. For example, some anticoagulants are contraindicated in patients with active massive bleeding, and certain sensitizing drugs are contraindicated in individuals with allergies. This invention does not limit the specific content of clinical decision-making pathways and drug contraindications.

[0046] S1023. Concatenate the clinical decision-making path, drug contraindications, user location information, and environmental information from the structured consultation data to construct a multi-source reasoning context.

[0047] After obtaining the clinical decision-making pathway and drug contraindications, the lightweight medical consultation AI model is then personalized by combining the specific circumstances of the user's situation. Specifically, the retrieved clinical decision-making pathway and drug contraindications are combined with the user's location information and environmental information (such as being in the field, without a hospital, or in high temperature) from the structured consultation data to form a multi-source reasoning context containing multi-dimensional information, enabling subsequent outputs to adapt to the on-site conditions.

[0048] S1024. Based on the multi-source reasoning context, perform multi-source information fusion reasoning to generate medical guidance text including personalized triage suggestions, environment-adaptive emergency operation steps, and medication contraindication reminders.

[0049] The lightweight medical consultation AI model takes multi-source reasoning context as input, performs multi-source information fusion reasoning, and comprehensively considers the general constraints of clinical decision-making paths and the specific limitations of the on-site environment to generate final medical guidance text that includes personalized triage suggestions, environment-adaptive emergency operation procedures, and medication contraindication reminders. For example, when the symptom is massive active bleeding, the environment is in the wild, and only a triangular bandage is available, the model outputs specific operation procedures such as applying pressure bandage, elevating the affected limb, and loosening it at regular intervals, as well as contraindication reminders such as not using hemostatic powder.

[0050] The spaceborne AI emergency medical consultation method provided by this invention transforms symptoms into numerical vectors through semantic vectorization and performs similarity matching in a spaceborne emergency medical knowledge base. This overcomes the limitations of traditional keyword matching and improves the accuracy of clinical decision path retrieval. Simultaneously, the retrieved decision path is concatenated with user location and environmental information to form a multi-source inference context, enabling the generated emergency guidance to adapt to different on-site resources and weather conditions, enhancing its practicality and relevance. The entire inference process is completed on-board, without ground intervention, protecting user privacy and compressing the response time of personalized emergency guidance within the satellite's transit cycle. This provides crucial time for self-rescue and mutual aid in emergency medical scenarios, effectively addressing the limitations of existing solutions in handling verbalized injuries and providing environmentally adaptive guidance.

[0051] In some embodiments, the satellite may receive medical consultation short messages from multiple user terminals within the same transit cycle. To ensure that high-risk emergencies can be given priority treatment, the onboard AI emergency medical consultation method provided by the present invention also includes a priority scheduling mechanism.

[0052] For example, before calling the lightweight medical consultation AI model, the symptom features in the structured consultation data can first be matched with keywords indicating the degree of urgency. For instance, preset high-risk emergency keywords can include, but are not limited to, one or more of the following: coma, cardiac arrest, uncontrollable bleeding, difficulty breathing, and severe pain. When any of these keywords is matched in the structured consultation data, a corresponding high-priority tag is generated for the current medical consultation message. Based on this high-priority tag, the current structured consultation data is prioritized, meaning it is placed at the front of the processing queue, allowing the lightweight medical consultation AI model to prioritize decision-making and inference based on this current structured consultation data, generating medical guidance text.

[0053] Optionally, if multiple high-priority consultations exist simultaneously, they can be further sorted by the number of matched keywords or their severity. If the current lightweight medical emergency AI model is processing low-priority consultations, the task can be interrupted or temporarily stored, and high-priority consultations can be processed instead.

[0054] Through the aforementioned priority scheduling steps, this invention can ensure that the most urgent medical consultations receive the fastest response even with limited on-board computing resources, thus maximizing the golden treatment time for critically ill patients.

[0055] S103. Encode the medical guidance text into a short reply message and send it to the user terminal via the downlink.

[0056] After generating the medical guidance text, it can be encapsulated into a format suitable for satellite-to-ground short message communication. For example, the satellite encodes the medical guidance text, including personalized triage suggestions, environmentally adapted emergency procedures, and medication contraindications, into one or more binary or text short messages according to a preset short message protocol. After encoding, the satellite directly sends the reply short message to the user terminal within its coverage area via its downlink, such as the UHF (Ultra High Frequency) band, L-band, or IoT communication band. The entire transmission process does not rely on ground gateway stations or public network facilities, achieving point-to-point transmission from satellite to terminal. After receiving the short message, the user terminal automatically decodes and displays the medical guidance text, allowing the user to perform self-rescue or mutual rescue based on the guidance.

[0057] In encoding medical guidance text, the bandwidth limitations of the satellite-to-ground link and the decoding capabilities of the user terminal can also be considered to ensure that complete medical guidance information is contained within a limited data length.

[0058] Through the aforementioned on-board real-time encoding and direct transmission, this invention can ensure that the entire process from receiving the consultation to delivering the guidance is completed within the satellite's transit period, meeting the timeliness requirements of emergency medical care.

[0059] In some embodiments, the spaceborne AI emergency medical consultation method provided by the present invention further includes the step of storing consultation record data in an on-board non-volatile memory and transmitting it to a ground system when appropriate.

[0060] For example, after receiving a short reply message, the satellite packages the original short message (medical consultation message), the generated medical guidance text, and the processing timestamp into a complete consultation record data. This data packet is encrypted and written to the onboard non-volatile memory to prevent loss due to power failure or space radiation. When the satellite reaches the ground station's communication coverage area according to its predetermined orbit, it automatically establishes a data transmission link with the ground station and downloads the stored consultation record data in batches to the ground rescue system. After receiving the data, the ground rescue system can use it for medical dispatch (such as coordinating subsequent professional rescue), medical record archiving, model training data supplementation, and public health monitoring and other management needs.

[0061] As can be seen, through this data storage and download mechanism, the spaceborne AI emergency medical consultation provided by this invention can not only ensure that users can obtain real-time medical guidance from the satellite in areas without terrestrial network coverage, but also ensure that ground rescue forces can obtain complete consultation records, thus achieving coordination between autonomous response from the satellite and global dispatch from the ground.

[0062] The satellite-borne AI emergency medical consultation method provided by this invention is applied to satellites. First, it receives short medical consultation messages containing colloquial descriptions of injuries sent by user terminals. It then analyzes and extracts symptom features, location information, and environmental information to generate structured consultation data adapted to clinical triage logic. Next, it invokes a lightweight medical consultation AI model deployed on the satellite to perform decision-making reasoning and generate medical guidance text. Finally, it encodes the text into a reply message and sends it to the user terminal via downlink. This invention executes the entire process of colloquial understanding, clinical structured conversion, and medical reasoning on the satellite end, eliminating the need for ground stations and data centers. This reduces consultation response time from tens of minutes to several hours to within the satellite's transit cycle, meeting the critical window for emergency rescue. Furthermore, the onboard AI model achieves accurate understanding of colloquial descriptions of injuries and automatic mapping to clinical triage logic, handling highly specialized issues such as symptom diversity and drug contraindications. It effectively solves the fundamental deficiency of existing systems that can only forward coordinates and simple text and lack medical semantic understanding capabilities, preventing trapped individuals from aggravating their injuries due to incorrect self-rescue attempts. Meanwhile, the entire consultation interaction is completed in a closed loop on the satellite, without relying on ground medical resources and communication networks. It can still operate autonomously in extreme scenarios such as damage to ground facilities or communication interruption, providing immediate, professional and reliable emergency medical support for areas without ground network coverage.

[0063] In some embodiments, the user's injury may change over time or require additional description. Satellite transit windows may also limit the time required to complete a full consultation through multiple transits. Therefore, the onboard AI emergency medical consultation provided by this invention also includes, for example... Figure 3 The multi-round consultation mechanism is shown. Figure 3 This is the third flowchart illustrating the spaceborne AI emergency medical consultation method provided by the present invention, as shown below. Figure 3As shown, it includes: S201. When a subsequent medical consultation short message is received from the user terminal, the consultation dialogue history of the same user terminal stored by the satellite during multiple transit cycles is obtained.

[0064] S202. Parse the subsequent medical consultation short messages to generate subsequent structured consultation data.

[0065] S203. Combine the subsequent structured consultation data with the consultation dialogue history to construct a multi-round consultation context.

[0066] S204. Call the lightweight medical consultation AI model to make decision-making inferences based on multi-round consultation context and generate continuous medical guidance text.

[0067] S205. Encode the continuous medical guidance text into a reply short message and send it to the user terminal via the downlink to respond to subsequent medical consultation short messages.

[0068] When a satellite receives a subsequent medical consultation message from the same user terminal during a later transit cycle, it first retrieves the user terminal's consultation dialogue history stored by the satellite over multiple transit cycles. This consultation dialogue history includes structured consultation data parsed from previous transits, medical guidance text generated by a lightweight medical consultation AI model, and timestamps. The satellite then parses the subsequent medical consultation message, extracting symptom features, user location information, and environmental information to generate structured consultation data for the subsequent medical consultation message, i.e., subsequent structured consultation data. Next, the subsequent structured consultation data is concatenated with the historical consultation dialogue history, linking new and old symptoms, disease progression, and implemented emergency measures in chronological order to form a multi-round consultation context. Then, the lightweight medical consultation AI model is invoked to perform decision-making reasoning based on this multi-round consultation context, comprehensively considering changes in the condition and the effectiveness of previous guidance, to generate continuous medical guidance text. For example, continuous medical guidance text might be: "Based on your previous description of the fracture, the swelling has worsened. Please continue to elevate the affected limb and apply ice." Finally, the continuous medical guidance text is encoded into a short response message and sent to the user terminal via the downlink to respond to the subsequent consultation.

[0069] Through this multi-round consultation mechanism, the spaceborne AI emergency medical consultation method provided by this invention can achieve continuous medical guidance across multiple satellite passes, effectively tracking the dynamic changes in the user's injury, avoiding repeated descriptions of medical history, and providing progressive and personalized emergency advice based on the initial consultation. This mechanism supports step-by-step descriptions of complex injuries and subsequent follow-up questions, enabling the spaceborne AI model to conduct patient follow-up and efficacy evaluation like a doctor, greatly improving the interactive depth and guidance accuracy of emergency medical consultations. Furthermore, since the entire consultation history is stored on the satellite, without relying on terrestrial networks, complete closed-loop management of the patient's condition can be achieved even in remote areas without terrestrial coverage, securing continuous and professional medical support for long-term rescue efforts.

[0070] Figure 4 This diagram illustrates the timing comparison between the spaceborne AI emergency medical consultation method provided by this invention and the traditional store-and-forward mode. It compares the differences in response time between the traditional store-and-forward mode and the spaceborne AI real-time consultation mode implemented by the spaceborne AI emergency medical consultation method provided by this invention. Figure 4 As shown in the upper part, in the traditional store-and-forward mode, after a user terminal sends a distress or medical inquiry short message, the IoT satellite needs to wait for the satellite to pass over the ground station before transmitting the data to the ground station. After processing by the data center or core network, the response is then transmitted back to the user terminal via the satellite link. The entire process involves the sum of the satellite transit period and the ground processing delay, and the response time is usually from tens of minutes to several hours. Figure 4 As shown in the lower half, in the onboard AI real-time consultation mode of this invention, after the user terminal sends a consultation short message, the IoT satellite immediately activates the onboard AI consultation module during its transit coverage period. The module completes short message parsing, semantic understanding, structured consultation data generation, decision reasoning, and medical guidance text generation onboard, without waiting for a ground station or relying on a data center. The response short message is directly sent to the user terminal within this transit cycle, with the response time only constrained by the satellite transit cycle, for example, 10-30 minutes. This timeline comparison intuitively demonstrates the significant improvement in emergency medical consultation response speed and the onboard autonomous closed-loop processing capability that completely eliminates reliance on ground facilities.

[0071] The following three application examples illustrate the spaceborne AI emergency medical consultation method provided by this invention. It should be understood that the following examples are only used to more clearly demonstrate the processing flow and interaction effects of this invention in actual emergency medical scenarios, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can make adaptive adjustments to the examples according to the actual spaceborne computing power, communication system, and terminal type. All equivalent substitutions or improvements made based on the core concept of this invention fall within the scope of protection of this invention.

[0072] Example 1 illustrates the first aid for a single person with a fracture in the wild: A geological prospector fell in an uninhabited area, resulting in a closed fracture of his right leg. He sent a short medical consultation message via a handheld terminal, for example: "Severe pain in my right calf, unable to stand, suspected fracture, no one around, how to immobilize it?" Upon receiving the medical consultation message via satellite overpass, the system first analyzes it, extracting symptom characteristics (such as right calf pain, inability to stand, suspected fracture), user location information, and environmental information (such as an uninhabited area or the absence of others), generating structured consultation data adapted to clinical triage logic. Subsequently, a lightweight medical consultation AI model is invoked for decision-making reasoning. This model matches the symptom characteristics with fracture first aid templates in the onboard emergency medical knowledge base, combining environmental information such as the absence of others and medical resources to generate medical guidance text. This medical guidance text might be: "Please keep the injured leg immobile. Find hard objects such as branches or hiking poles and place them on both sides of the calf. Secure them with shoelaces or strips of clothing, placing the securing points above and below the fracture to avoid compressing the fracture site. After immobilization, elevate the affected limb to reduce swelling." The system will continue to monitor your condition during satellite overpasses. The satellite encodes the text into a short reply message and sends it to the user terminal via the downlink. Upon receiving the reply message, the user follows the instructions to complete the fixation process, thus preventing secondary damage.

[0073] Example 2 illustrates the case of mass consultations in a multi-person poisoning incident: A scientific expedition team accidentally ingested poisonous mushrooms in a remote mountainous area. Several members experienced vomiting and diarrhea. The team leader sent a short medical consultation message via a terminal, for example: "Six people have ingested poisonous mushrooms two hours ago, all experiencing vomiting and diarrhea. There is no hospital available. What should be done?" After the satellite receives the message, it analyzes it to extract symptom characteristics (such as vomiting, diarrhea, and the fact that the mushrooms were ingested two hours ago), user location information (such as coordinates provided by the terminal or the location in the mountainous area extracted from the message), and environmental information (such as remote mountainous area and lack of hospital access), thereby generating structured consultation data. A lightweight medical consultation AI model, based on the typical symptoms of mushroom poisoning, retrieves clinical decision-making pathways such as inducing vomiting, fluid replacement, and sample collection from an emergency medical knowledge base. Combining this with the on-site environment of no hospital access, it generates medical guidance text, for example: "Please immediately stop eating. Each person should drink plenty of warm water and then try to induce vomiting. Preserve a sample of the vomit. If confusion or difficulty breathing occurs, prioritize the treatment of the most severely ill. Maintain communication. The satellite will then pass over the satellite every few tens of minutes to monitor changes in the condition." The satellite will then send a reply message to the user's terminal. Following instructions, the team leader organized induced vomiting and nursing care to stabilize the condition of all personnel until professional rescue arrived.

[0074] Example 3 illustrates the situation using the failure of a full-surface network under extreme disaster conditions: In an earthquake-stricken area where all ground base stations were destroyed, affected residents sent numerous short medical consultation messages via emergency terminals, covering topics such as external injuries, crush injuries, and dehydration. During its transit, the satellite received all consultation requests, parsed each message sequentially, generated structured consultation data, and invoked a lightweight medical consultation AI model to perform on-orbit decision-making. It then generated corresponding medical guidance texts for different injuries, such as hemostasis, wound bandaging, finding safe water sources, and preventing crush syndrome, and immediately replied to each user terminal via downlink. All consultation records were simultaneously stored in the satellite's non-volatile memory. This process requires no involvement from ground stations or data centers, effectively reducing secondary casualties after the disaster. Once ground communication was restored, the satellite batch-downloaded historical consultation records to the ground rescue system, providing data support for subsequent precise rescue and dispatch.

[0075] The above embodiments correspond to single-person fracture emergency care, mass poisoning incidents, and batch consultation scenarios under extreme disasters, respectively. They demonstrate the entire process of the spaceborne AI emergency medical consultation method provided by the present invention, from receiving, parsing and structuring consultation short messages, AI decision-making reasoning, generating guidance text, to issuing responses. This can verify the immediacy, professionalism, and reliability of the present invention in areas without terrestrial network coverage.

[0076] The following describes the spaceborne AI emergency medical consultation device provided by the present invention. The spaceborne AI emergency medical consultation device described below can be referred to in correspondence with the spaceborne AI emergency medical consultation method described above.

[0077] Figure 5 This is a schematic diagram of the structure of the spaceborne AI emergency medical consultation device provided by the present invention, which is applied to a satellite. Figure 5 As shown, the spaceborne AI emergency medical consultation device 400 provided by the present invention includes: The data receiving and parsing module 401 is used to receive medical consultation short messages containing colloquial descriptions of injuries sent by user terminals, and to parse the medical consultation short messages to extract symptom features, user location information and environmental information, and generate structured consultation data that is adapted to clinical triage logic. The decision reasoning module 402 is used to call the lightweight medical consultation AI model deployed on the satellite, perform decision reasoning based on structured consultation data, and generate medical guidance text; The text delivery module 403 is used to encode medical guidance text into a short reply message and send it to the user terminal via the downlink.

[0078] In one possible design, the data receiving and parsing module 401 is used for: The system performs text cleaning and semantic segmentation on short medical consultation messages to identify symptom features, user location information, and environmental information. After normalizing and aligning the identified information, it is mapped to predefined standardized fields to generate structured consultation data that adapts to clinical triage logic.

[0079] In one possible design, the decision reasoning module 402 is used for: The symptom features in structured medical consultation data are semantically vectorized to generate symptom semantic vectors; Using symptom semantic vectors as search criteria, similarity matching is performed in the spaceborne emergency medical knowledge base to retrieve associated clinical decision-making paths and drug contraindications. By concatenating the clinical decision-making pathway, drug contraindications, and user location and environmental information from structured consultation data, a multi-source reasoning context is constructed. Based on multi-source reasoning context, multi-source information fusion reasoning is performed to generate medical guidance texts that include personalized triage suggestions, environment-adaptive emergency operation procedures, and medication contraindication reminders.

[0080] In one possible design, the decision reasoning module 402 is also used for: The structured consultation data is matched with keywords indicating the urgency level. If a preset high-risk emergency keyword is matched, a high-priority marker is generated for the current medical consultation short message. Based on high-priority labeling, the current structured consultation data is prioritized for scheduling, enabling the lightweight medical consultation AI model to prioritize decision-making and reasoning based on the current structured consultation data.

[0081] In one possible design, the decision reasoning module 402 is also used for: When a subsequent medical consultation short message is received from the user terminal, the consultation dialogue history of the same user terminal stored by the satellite in multiple transit cycles is obtained. The system parses subsequent medical consultation short messages to generate subsequent structured consultation data. The subsequent structured consultation data is spliced ​​with the consultation dialogue history to construct a multi-round consultation context; A lightweight medical consultation AI model is invoked to make decision-making inferences based on multi-round consultation contexts and generate continuous medical guidance texts. The continuous medical guidance text is encoded into a short response message and sent to the user terminal via the downlink to respond to subsequent medical consultation short messages.

[0082] In one possible design, the text delivery module 403 is also used for: The medical consultation short message, medical guidance text and processing timestamp are packaged into consultation record data and written into the onboard non-volatile memory; When the satellite flies over the ground station's communication coverage area, it will download the consultation record data in batches to the ground rescue system for use in medical dispatch and medical record archiving.

[0083] In one possible design, the lightweight medical consultation AI model is trained based on large-scale clinical dialogue data, emergency guidelines, and a drug contraindication knowledge base, and then subjected to model pruning, quantization, or knowledge distillation and compression to adapt to the computing power and storage constraints of satellite embedded processors.

[0084] Figure 6 This is a schematic diagram of the structure of the IoT satellite provided by the present invention, such as... Figure 6 As shown, the IoT satellite may include: an on-board processor 610, a space-ground communication interface 620, an on-board memory 630, and an internal data bus 640. The on-board processor 610, the space-ground communication interface 620, and the on-board memory 630 communicate and transmit commands to each other via the internal data bus 640. The on-board processor 610 can call logical instructions in the on-board memory 630 to execute an on-board AI emergency medical consultation method. This method, applied to the satellite, includes: receiving a short medical consultation message containing a colloquial description of the injury sent by a user terminal; parsing the message to extract symptom features, user location information, and environmental information; generating structured consultation data adapted to clinical triage logic; calling a lightweight medical consultation AI model deployed on the satellite; performing decision-making reasoning based on the structured consultation data; and encoding the medical guidance text into a reply message and sending it to the user terminal via a downlink.

[0085] Furthermore, the logical instructions in the aforementioned onboard memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an onboard computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks or optical disks, and onboard non-volatile memory.

[0086] The present invention also provides a spaceborne AI emergency medical consultation system, including: a user terminal and the aforementioned IoT satellite; The user terminal communicates with the IoT satellite via a satellite-to-ground link to send short messages for medical consultations and receive short reply messages from the IoT satellite.

[0087] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the satellite-borne AI emergency medical consultation method provided by the above methods. This method is applied to a satellite and includes: receiving a medical consultation short message containing a colloquial description of the injury sent by a user terminal, parsing the medical consultation short message, extracting symptom features, user location information, and environmental information, and generating structured consultation data adapted to clinical triage logic; calling a lightweight medical consultation AI model deployed on the satellite, performing decision reasoning based on the structured consultation data, and generating medical guidance text; encoding the medical guidance text into a reply short message, and sending it to the user terminal via a downlink.

[0088] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program is implemented to perform the satellite-borne AI emergency medical consultation method provided by the above methods. This method is applied to a satellite and includes: receiving a short medical consultation message containing a colloquial description of the injury sent by a user terminal, parsing the short medical consultation message, extracting symptom features, user location information, and environmental information, and generating structured consultation data adapted to clinical triage logic; calling a lightweight medical consultation AI model deployed on the satellite, performing decision reasoning based on the structured consultation data, and generating medical guidance text; encoding the medical guidance text into a reply short message, and sending it to the user terminal via a downlink.

[0089] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0090] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A spaceborne AI-based emergency medical consultation method, characterized in that, Applied to satellites, including: The system receives short medical consultation messages containing colloquial descriptions of injuries sent by user terminals, parses the short medical consultation messages, extracts symptom features, user location information, and environmental information, and generates structured consultation data that adapts to clinical triage logic. The lightweight medical consultation AI model deployed on the satellite is invoked to perform decision-making reasoning based on the structured consultation data and generate medical guidance text. The medical guidance text is encoded into a short reply message and sent to the user terminal via a downlink.

2. The method according to claim 1, characterized in that, The process of parsing the medical consultation short message, extracting symptom features, user location information, and environmental information, and generating structured consultation data adapted to clinical triage logic includes: The medical consultation short message is cleaned and semantically segmented to identify the symptom features, user location information, and environmental information. After the identified information is normalized and formatted, it is mapped to predefined standardized fields to generate the structured consultation data adapted to the clinical triage logic.

3. The method according to claim 1, characterized in that, The process of calling upon the lightweight medical consultation AI model deployed on the satellite to perform decision-making reasoning based on the structured consultation data and generate medical guidance text includes calling upon the lightweight medical consultation AI model to perform the following steps: The symptom features in the structured consultation data are semantically vectorized to generate symptom semantic vectors; Using the symptom semantic vector as the retrieval criteria, similarity matching is performed in the spaceborne emergency medical knowledge base to retrieve the associated clinical decision-making paths and drug contraindications. The clinical decision-making path, the drug contraindications, the user location information, and the environmental information in the structured consultation data are concatenated to construct a multi-source reasoning context; Based on the multi-source reasoning context, multi-source information fusion reasoning is performed to generate the medical guidance text, which includes personalized triage suggestions, environment-adaptive emergency operation steps, and medication contraindication reminders.

4. The method according to claim 1, characterized in that, Prior to invoking the lightweight medical consultation AI model deployed on the satellite, the following is also included: The structured consultation data is matched with keywords indicating urgency. If a preset high-risk emergency keyword is matched, a high-priority marker is generated for the current medical consultation short message. Based on the high-priority label, the current structured consultation data is prioritized for scheduling, enabling the lightweight medical consultation AI model to prioritize decision-making and reasoning based on the current structured consultation data.

5. The method according to claim 1, characterized in that, Also includes: When a subsequent medical consultation short message is received from the user terminal, the consultation dialogue history of the same user terminal stored by the satellite during multiple transit cycles is obtained. The subsequent medical consultation short messages are parsed to generate subsequent structured consultation data; The subsequent structured consultation data is concatenated with the consultation dialogue history to construct a multi-round consultation context; The lightweight medical consultation AI model is invoked to perform decision-making reasoning based on the multi-round consultation context and generate continuous medical guidance text. The continuous medical guidance text is encoded into the reply short message and sent to the user terminal via the downlink to respond to the subsequent medical consultation short message.

6. The method according to claim 1, characterized in that, Also includes: The medical consultation short message, the medical guidance text, and the processing timestamp are packaged into consultation record data and written into the onboard non-volatile memory. When the satellite flies over the communication coverage area of ​​the ground station, it will download the consultation record data in batches to the ground rescue system for medical dispatch and medical record archiving.

7. The method according to any one of claims 1 to 6, characterized in that, The lightweight medical consultation AI model is trained based on large-scale clinical dialogue data, emergency guidelines, and a drug contraindication knowledge base. It is then subjected to model pruning, quantization, or knowledge distillation compression to adapt to the computing power and storage constraints of the satellite embedded processor.

8. A spaceborne AI emergency medical consultation device, characterized in that, Applied to satellites, including: The data receiving and parsing module is used to receive medical consultation short messages containing colloquial descriptions of injuries sent by user terminals, and to parse the medical consultation short messages to extract symptom features, user location information and environmental information, and generate structured consultation data that is adapted to clinical triage logic. The decision reasoning module is used to call the lightweight medical consultation AI model deployed on the satellite, perform decision reasoning based on the structured consultation data, and generate medical guidance text; The text delivery module is used to encode the medical guidance text into a short reply message and send it to the user terminal via a downlink.

9. An Internet of Things (IoT) satellite, characterized in that, include: The satellite-borne memory, the satellite-borne processor, and the computer program stored on the satellite-borne memory and running on the satellite-borne processor, wherein the satellite-borne processor, when executing the computer program, implements the satellite-borne AI emergency medical consultation method as described in any one of claims 1 to 7.

10. A spaceborne AI emergency medical consultation system, characterized in that, include: User terminal and IoT satellite as described in claim 9; The user terminal is connected to the IoT satellite via a satellite-to-ground link for sending short messages for medical consultations and receiving short reply messages from the IoT satellite.

Images