CN-122025102-A - Multimode fusion type auxiliary system for first-aid AI (advanced treatment) before animal injury

CN122025102ACN 122025102 ACN122025102 ACN 122025102ACN-122025102-A

Abstract

The invention relates to a multimode fusion type auxiliary system for first-aid AI before animal injury. The technical scheme includes that the intelligent animal emergency system comprises a multi-mode data acquisition module, a data preprocessing and fusion module, an AI intelligent analysis decision module, an emergency instruction interaction module and a data storage and tracing module, wherein the modules are in bidirectional linkage through a wireless communication link, adapt to various pre-hospital emergency scenes of a mobile phone, an emergency terminal and intelligent wearing equipment, and realize full-flow closed-loop assistance from data acquisition to emergency instruction of animal injury. The invention has the beneficial effects of solving the problem of inaccurate judgment of the first-aid injury condition before the animal is injured, standardizing the operation flow of the first-aid operation before the hospital, shortening the connection time of the medical empty window period and the first-aid treatment, supporting various terminal equipment and different scenes, meeting various use requirements of mobile phones, first-aid terminals and the like, providing reliable data support for first-aid quality assessment, model optimization and medical dispute treatment, protecting the privacy of the wounded through encryption treatment, and taking the practicality and the safety into consideration.

Inventors

ZHANG YANQUN
LI JUNJIE
CHEN JIJUN
Dang Yuantao
LI BOLONG

Assignees

中国人民解放军空军军医大学

Dates

Publication Date: 20260512
Application Date: 20260131

Claims (8)

1. The multimode fused animal pre-hospital emergency AI auxiliary system is characterized by comprising a multimode data acquisition module, a data preprocessing and fusion module, an AI intelligent analysis decision module, an emergency guidance interaction module and a data storage and tracing module, wherein the modules are in bidirectional linkage through a wireless communication link, adapt to various pre-hospital emergency scenes of a mobile phone, an emergency terminal and intelligent wearing equipment, and realize full-flow closed-loop auxiliary from data acquisition to emergency guidance of animal injuries.
2. The multimode fused animal pre-hospital injury emergency AI auxiliary system of claim 1 is characterized in that the multimode data acquisition module automatically acquires three types of core data, namely, images of wounds and injured animals are acquired through a terminal camera, voice data of injuries and injured processes are acquired through a built-in microphone, dialect and fuzzy voice analysis is supported, heart rate and blood pressure physiological sign data of the injured are acquired through a docking intelligent device, abnormal data are filtered, and the acquisition frequency is adjusted in a self-adaptive mode.
3. The multi-modal fusion pre-hospital-injury first-aid AI auxiliary system of claim 1, wherein the data preprocessing and fusion module performs targeted preprocessing on raw data of each mode, and generates a unified injury characteristic vector by combining low-level characteristic splicing and high-level semantic fusion, so that AI analysis accuracy is improved, and judgment errors of single-mode data loss are reduced.
4. The multi-modal fusion pre-hospital emergency AI auxiliary system for animal injury is characterized in that an AI intelligent analysis decision module is internally provided with a trained and optimized multi-task identification model, based on an animal injury clinical case library, an emergency guide and a multi-modal training data set, multi-dimensional synchronous analysis is realized, firstly, the type identification of the injured animal is combined with collected animal image characteristics and voice description, specific types of the injured animal are identified, the frequently injured animal such as dogs, cats, venomous snakes and bees is distinguished, the dangerous grade of the animal and the pathogenic microorganisms and toxin types possibly carried by the animal are marked, secondly, the injury is classified and judged in a grading manner, the injury is classified into four grades of mild, moderate, severe and critical degree by combining the fused injury characteristic vectors, the judgment basis of the injury at each grade is defined, the emergency of hemorrhagic shock and toxin poisoning is identified, thirdly, an emergency guide scheme is generated according to the types of the injured animal, the injury grading and physiological signs of the injured person, and the scheme covers wound treatment, hemostasis, assistance, body position adjustment specific operation steps and special operation contraindication are carried out, and medical equipment is suitable for the medical equipment before the medical treatment is marked.
5. The multi-mode fused animal pre-hospital injury emergency AI auxiliary system as set forth in claim 1, wherein the emergency guidance interaction module presents emergency operations step by step through graphics and texts and voice, supports operation progress feedback and warning reminding, and can manually supplement injury information and update a scheme synchronously to realize real-time data linkage with a near-hospital.
6. The multi-mode fused pre-hospital injury emergency AI auxiliary system of claim 1 wherein the data storage and traceability module is in an encrypted storage mode, stores data related to whole-process emergency and comprises a timestamp and an identity, supports local caching and cloud synchronization, has a data traceability function, and can inquire emergency data according to keywords.
7. The multi-modal fusion pre-hospital injury emergency AI assistance system as set forth in claim 4 wherein said AI intelligent analysis decision module automatically collects emergency data and clinical cases for model incremental training with self-learning and adaptive optimization capabilities and adjusts analysis parameters and protocol emphasis based on different regional characteristics.
8. The multi-mode fused pre-hospital injury emergency AI auxiliary system of any one of claims 1-7 further comprising emergency positioning linkage and privacy protection functions, wherein an offline operation mode is supported, a core emergency plan can be cached, and basic emergency guidance requirements in a network-free scene are met.

Description

Multimode fusion type auxiliary system for first-aid AI (advanced treatment) before animal injury Technical Field The invention belongs to the technical field of AI auxiliary first aid, and relates to a multimode fusion type AI auxiliary system for first aid before animal injury. Background Animal injury is a common emergency in pre-hospital first aid, and comprises dog bite, cat scratch, venomous snake bite, bee sting and the like, if the injury is not timely and standard in treatment, serious complications such as wound infection, toxin diffusion, hemorrhagic shock and the like are easily caused, and even the lives of wounded persons are endangered. At present, the pre-hospital emergency link for animals has a plurality of pain points, namely, firstly, the difficulty of judging the injury is large, most of pre-hospital emergency personnel are non-professional medical personnel, the types of the injured animals, the severity degree of the injury and the toxin type cannot be judged quickly and accurately, misjudgment and missed judgment are easy to occur, secondly, the treatment process is not standard, the emergency levels of different emergency personnel are uneven, unified and scientific emergency guidance is lacking, the problems of improper hemostasis and bandaging, incorrect detoxification operation and the like are easy to occur, the optimal treatment opportunity is delayed, thirdly, the medical air window period is long, the number of pre-hospital emergency scenes is large, the emergency personnel and the medical personnel in the hospital lack of effective linkage, the injured data cannot be synchronized timely, the in-hospital treatment preparation is insufficient, the linking efficiency is low, and fourthly, the traditional emergency auxiliary tool is single in mode, only can provide basic emergency knowledge inquiry or single data acquisition function, the comprehensive analysis of multi-source injury related data cannot be integrated, the auxiliary decision accuracy is low, and the requirement of the pre-hospital emergency scene is difficult to meet. Therefore, the invention provides a multimode fusion type animal pre-hospital injury first-aid AI auxiliary system, which improves the accuracy, normalization and high efficiency of animal pre-hospital injury first-aid, shortens the treatment connection time, reduces the complication occurrence rate and solves the problems through a multimode data fusion technology. Disclosure of Invention In view of the problems existing in the prior art, the invention discloses a multimode fusion type animal pre-hospital emergency AI auxiliary system, which adopts the technical scheme that the multimode fusion type animal pre-hospital emergency AI auxiliary system comprises a multimode data acquisition module, a data preprocessing and fusion module, an AI intelligent analysis decision module, an emergency guidance interaction module and a data storage and tracing module, wherein the modules are in bidirectional linkage through a wireless communication link, adapt to various pre-hospital emergency scenes of a mobile phone, an emergency terminal and intelligent wearing equipment, and realize full-flow closed-loop auxiliary from data acquisition to emergency guidance of animal injuries. The multimode data acquisition module adopts a multisource heterogeneous data synchronous acquisition mode to automatically acquire three types of core data, namely image data, high-definition images of wounds and appearance images of animals to be injured are acquired through a terminal camera, night light supplementing shooting and local amplification acquisition are supported, the breaking degree, bleeding state and animal species characteristics of the wounds are captured, voice data are acquired through an embedded microphone, the injury information and injury process described by emergency personnel or injured persons are supported, dialect recognition and fuzzy voice analysis are supported, physiological sign data are acquired through docking an intelligent wearable device or a portable monitor, abnormal condition data caused by environmental interference are synchronously filtered, and the acquisition frequency is adaptively adjusted according to the emergency degree of the injury. The data preprocessing and fusion module adopts a hierarchical fusion strategy, firstly carries out targeted preprocessing on original data of each mode, namely denoising, graying and feature extraction on image data, retaining key features of wound edges and bleeding areas, carrying out denoising, word segmentation and semantic conversion on voice data, converting voice information into structured text data, carrying out outlier rejection and normalization processing on physiological sign data, ensuring data stability, and fusing preprocessed multi-mode data in a mode of combining low-level feature splicing and high-level semantic fusion to generate a unified injury feature vector, improving the ac