Pig disease diagnosis knowledge base based on semantic segmentation and pig disease diagnosis method thereof

Abstract

The application provides an AI diagnosis pig disease knowledge base based on semantic segmentation and a pig disease diagnosis method thereof, and the system comprises a voice interaction module, a semantic segmentation module, a SOAP structured medical record generation module, a knowledge base module and a diagnosis reasoning module. A veterinarian can complete disease recording by speaking, the system automatically segments and quantifies lesion characteristics from images, uses a deep learning model to identify diseases, classify diseases and analyze complications after fusing multi-modal data, and finally automatically generates a SOAP medical record meeting international standards and recommends a treatment scheme. The application realizes the intelligentization, automation and paperless of the whole process of pig disease diagnosis, especially achieves the goal of zero manual medical record input, and significantly improves the diagnosis efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of veterinary testing technology, and in particular to an AI-based knowledge base for diagnosing swine diseases based on semantic segmentation and a method for diagnosing swine diseases. Background Technology

[0002] Pig farming is one of the core pillars of my country's animal husbandry industry. As the scale of farming continues to expand, the prevention and control of pig diseases has become a key link affecting farming efficiency and biosecurity.

[0003] Traditional diagnostic methods rely entirely on the individual veterinarian's experience, with a high degree of subjective judgment. This leads to significant discrepancies in diagnostic conclusions among veterinarians of varying skill levels and years of experience, resulting in low overall diagnostic efficiency and frequent misdiagnosis and missed diagnosis. In terms of medical record management, the industry currently predominantly uses manual typing and paper-based ledgers to record patient information. This process is cumbersome, time-consuming, and labor-intensive. Furthermore, the formats of medical records from different pig farms and veterinarians are inconsistent and disorganized, failing to create standardized, structured data that is insufficient for subsequent case review, disease tracing, and industry data statistical analysis.

[0004] As artificial intelligence technology gradually penetrates the livestock industry, some pig disease auxiliary diagnostic equipment and software based on ordinary image recognition have appeared on the market. However, existing product technical solutions have significant shortcomings: First, image analysis can only complete coarse-grained region recognition, lacking pixel-level semantic segmentation capabilities, making it impossible to accurately distinguish between healthy tissues and lesions in pigs, and difficult to quantify key features such as lesion location, area, shape, and color, resulting in insufficient lesion localization accuracy. Second, existing systems do not incorporate voice interaction technology, and disease information entry is still mainly manual, failing to fundamentally solve the problem of low efficiency in medical record entry. Third, various system functions are independent of each other, with modules such as image recognition, disease recording, diagnostic reasoning, and treatment plan recommendation operating in isolation. There is no unified industry knowledge base to integrate and apply professional knowledge such as historical cases, disease characteristics, medication guidelines, and complication patterns, resulting in weak intelligent reasoning capabilities. Fourth, most products only achieve single-diagnosis functions, lacking extended functions such as pre- and post-treatment lesion comparison analysis, disease progression prediction, breeding environment monitoring, and abnormal state early warning, failing to form a closed-loop management system covering the entire process from lesion discovery, diagnostic assessment, treatment intervention to prognosis tracking and risk warning, making it difficult to meet the intelligent epidemic prevention needs of modern large-scale pig farms. Summary of the Invention

[0005] This invention aims to provide an AI-based knowledge base for swine disease diagnosis based on semantic segmentation and its swine disease diagnosis method, so as to realize the intelligent, accurate and automated process of swine disease diagnosis. In particular, it can automatically generate SOAP structured medical records through voice interaction, achieving the goal of zero manual data entry.

[0006] One technical solution of the present invention is implemented as follows: A semantic segmentation-based AI-based knowledge base for diagnosing swine diseases includes: The voice interaction module is used to receive verbal information from veterinarians, parse it using a veterinary professional lexicon, and convert it into structured text. The image semantic segmentation module is used to perform pixel-level classification of pig images and accurately extract lesion region features; The SOAP medical record generation module is used to automatically generate standardized medical records containing four parts: subjective information, objective information, assessment, and plan, based on the analysis results of voice, text, and images. The knowledge base module is used to store historical cases, treatment plans, and drug information, and supports similar case retrieval and treatment recommendations; The disease diagnosis and assessment module, based on feature fusion and deep learning models, outputs disease category, disease grade, and complication risk. The treatment recommendation and early warning module is used to generate personalized treatment plans and monitor for abnormalities in real time to issue early warnings.

[0007] Preferably, the veterinary terminology database includes terms such as anatomical locations of pigs, disease names, symptom descriptions, drug names, and procedures, and supports user-defined expansion.

[0008] Preferably, the semantic segmentation module adopts a U-Net or DeepLabv3+ network structure to output the coordinates, area, shape, edge contour, and color anomaly annotation of the lesion.

[0009] Preferably, the SOAP medical record generation module requires zero manual input, and all fields are automatically filled by voice or image analysis.

[0010] Preferably, the disease diagnosis and assessment module integrates a time-series prediction model to predict disease progression, recovery period, and risk of deterioration.

[0011] Preferably, it also includes a real-time monitoring and early warning module, which simultaneously analyzes video images and environmental sensor data, identifies abnormal behavior and environmental exceedances through semantic segmentation, and pushes early warnings in a tiered manner.

[0012] Preferably, the knowledge base module supports reinforcement learning based on efficacy feedback, which can dynamically optimize the treatment plan recommendation strategy.

[0013] Another technical solution of the present invention is implemented as follows: AI-based swine disease diagnosis method based on semantic segmentation Includes the following steps: Step 1: The system receives the veterinarian's voice input, recognizes and parses it into structured subjective data; Step 2: Collect images of pigs and extract objective lesion features through semantic segmentation; Step 3: Integrate subjective and objective data, and call upon the AI ​​model for diagnostic assessment; Step 4: Search the knowledge base to generate treatment plans and nursing recommendations; Step 5: Automatically populate SOAP medical records and save them to the knowledge base; Step 6: Continuously monitor the pigs' condition and environment, trigger alerts, and update medical records.

[0014] By adopting the above technical solution, the beneficial effects of the present invention are as follows: This invention achieves zero manual data entry, significantly improving efficiency: through voice recognition, a professional lexicon, and SOAP templates, veterinarians can complete high-quality medical record entries without keyboard operation, reducing single-diagnosis time by more than 80%. Semantic segmentation technology is used for pixel-level lesion localization, combined with feature fusion and deep learning classification, achieving a disease identification accuracy rate of over 95%. Knowledge is reusable and evolving; the system's built-in knowledge base expands continuously with accumulated cases, and treatment plans can be continuously optimized based on efficacy feedback. From image acquisition, voice recording, lesion segmentation, automatic diagnosis, treatment recommendations to early warning monitoring, an integrated intelligent decision-making chain is formed. It lowers the medical threshold for the livestock industry: even non-professional veterinarians can use the system to complete preliminary diagnosis and treatment, making it particularly suitable for large-scale pig farms. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Example

[0016] One technical solution of the present invention is implemented as follows: A semantic segmentation-based AI-based knowledge base for diagnosing swine diseases includes: The voice interaction module is used to receive verbal information from veterinarians, parse it using a veterinary professional lexicon, and convert it into structured text. The image semantic segmentation module is used to perform pixel-level classification of pig images and accurately extract lesion region features; The SOAP medical record generation module is used to automatically generate standardized medical records containing four parts: subjective information, objective information, assessment, and plan, based on the analysis results of voice, text, and images. The knowledge base module is used to store historical cases, treatment plans, and drug information, and supports similar case retrieval and treatment recommendations; The disease diagnosis and assessment module, based on feature fusion and deep learning models, outputs disease category, disease grade, and complication risk. The treatment recommendation and early warning module is used to generate personalized treatment plans and monitor for abnormalities in real time to issue early warnings.

[0017] The veterinary terminology database includes terms such as anatomical locations of pigs, disease names, symptom descriptions, drug names, and manipulative actions, and supports user-defined expansion. The semantic segmentation module uses a U-Net or DeepLabv3+ network structure to output the coordinates, area, shape, edge contours, and color anomaly annotations of lesions. The SOAP medical record generation module requires zero manual input; all fields are automatically filled by voice or image analysis. The disease diagnosis and assessment module integrates a time-series prediction model to predict disease progression, recovery period, and risk of deterioration. A real-time monitoring and early warning module is also included. This module simultaneously analyzes video images and environmental sensor data, identifies behavioral abnormalities and environmental exceedances through semantic segmentation, and pushes tiered warnings.

[0018] The knowledge base module supports reinforcement learning based on efficacy feedback, which can dynamically optimize treatment plan recommendation strategies. Example

[0019] AI-based swine disease diagnosis method based on semantic segmentation Includes the following steps: Step 1: The system receives the veterinarian's voice input, recognizes and parses it into structured subjective data; Step 2: Collect images of pigs and extract objective lesion features through semantic segmentation; Step 3: Integrate subjective and objective data, and call upon the AI ​​model for diagnostic assessment; Step 4: Search the knowledge base to generate treatment plans and nursing recommendations; Step 5: Automatically populate SOAP medical records and save them to the knowledge base; Step 6: Continuously monitor the pigs' condition and environment, trigger alerts, and update medical records.

[0020] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A knowledge base for AI-based diagnosis of swine diseases based on semantic segmentation, characterized in that, include: The voice interaction module is used to receive verbal information from veterinarians, parse it using a veterinary professional lexicon, and convert it into structured text. The image semantic segmentation module is used to perform pixel-level classification of pig images and accurately extract lesion region features; The SOAP medical record generation module is used to automatically generate standardized medical records containing four parts: subjective information, objective information, assessment, and plan, based on the analysis results of voice, text, and images. The knowledge base module is used to store historical cases, treatment plans, and drug information, and supports similar case retrieval and treatment recommendations; The disease diagnosis and assessment module, based on feature fusion and deep learning models, outputs disease category, disease grade, and complication risk. The treatment recommendation and early warning module is used to generate personalized treatment plans and monitor for abnormalities in real time to issue early warnings.

2. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, The veterinary terminology database contains terms such as anatomical locations of pigs, disease names, symptom descriptions, drug names, and procedures, and supports user-defined expansions.

3. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, The semantic segmentation module uses a U-Net or DeepLabv3+ network structure to output the coordinates, area, shape, edge contour, and color anomaly annotations of the lesions.

4. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, The SOAP medical record generation module requires zero manual input; all fields are automatically filled by voice or image analysis.

5. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, The disease diagnosis and assessment module integrates a time-series prediction model to predict disease progression, recovery period, and risk of deterioration.

6. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, It also includes a real-time monitoring and early warning module, which simultaneously analyzes video images and environmental sensor data, identifies abnormal behavior and environmental exceedances through semantic segmentation, and pushes early warnings in a tiered manner.

7. The AI-based swine disease diagnostic knowledge base based on semantic segmentation according to claim 1, characterized in that, The knowledge base module supports reinforcement learning based on efficacy feedback, which can dynamically optimize treatment plan recommendation strategies.

8. The method for diagnosing swine diseases using the knowledge base described in any one of claims 1-7. Its features are, Includes the following steps: Step 1: The system receives the veterinarian's voice input, recognizes and parses it into structured subjective data; Step 2: Collect images of pigs and extract objective lesion features through semantic segmentation; Step 3: Integrate subjective and objective data, and call upon the AI ​​model for diagnostic assessment; Step 4: Search the knowledge base to generate treatment plans and nursing recommendations; Step 5: Automatically populate SOAP medical records and save them to the knowledge base; Step 6: Continuously monitor the pigs' condition and environment, trigger alerts, and update medical records.