PawCare+: AI-Powered Intelligent Pet Health Guidance System

Project Basic Information

• Original Author/Maintainer: Kiran7112
• Source Platform: GitHub
• Original Project Name: PawCare---AI-Pet-Health-Guidance-System
• Original Link: https://github.com/Kiran7112/PawCare---AI-Pet-Health-Guidance-System
• Release Time: June 13, 2026

Core Points

PawCare+ is a pet health guidance system that integrates machine learning prediction and large language model expertise. It offers pet owners personalized health management plans via 17 pet profile data extraction, health risk assessment, and care ability analysis.

Modern pet owners face common dilemmas: when their pets show abnormalities, online advice is one-size-fits-all, and consulting a vet is expensive; most online resources cannot provide personalized guidance based on the pet's specific situation, and there is a lack of targeted plans in emergencies.

PawCare+ was created to address this pain point. It integrates machine learning and large language model expertise to build an intelligent system that evaluates both pet health risks and owners' care abilities, making professional-level guidance accessible.

PawCare+ adopts a front-end and back-end separation architecture:

Frontend Layer

Built with React+TypeScript, using Vite as the build tool and Tailwind CSS for style management, it provides a responsive dashboard and interactive visualization interface.

Backend Layer

FastAPI serves as the REST API bridge, and Uvicorn provides asynchronous processing capabilities.

Intelligent Orchestration Layer

The core is a 17-node state machine workflow built with LangGraph, coordinating 16 intelligent agents, 2 machine learning models, and data integration nodes.

Model Layer

• Large Language Model: OpenAI GPT-4 for natural language understanding and generation
• Machine Learning Models: scikit-learn-trained health risk prediction and care ability scoring models
• Data Processing: pandas and numpy for data cleaning and feature engineering

The workflow starts with filling out an assessment form, collecting 17-dimensional pet profile information (type, breed, age, etc.):

Input Validation

Check data completeness and validity, and identify the pet type.

Path Diversion

Based on assessment results, divert to three paths:

• Emergency care path (high risk: emergency guidance + medical advice)
• Preventive care path (medium risk: daily monitoring + preventive measures)
• Health maintenance path (low risk: long-term management plan)

Multi-dimensional Guidance

Covers 5+ health areas including emergency response, nutrition management, and behavior training, providing AI-generated personalized suggestions.

Case: A 7-year-old overweight Labrador retriever shows increased thirst and drowsiness. The system routes it to the preventive care path, with a health risk of 39.1% (medium) and the owner's care ability score of 62.

Assessment results are presented via four tabs:

• Overview page: Care path banner + risk/ability dashboard
• Health guidance page: Personalized AI suggestions
• Detailed analysis page: Pet profile + machine learning metrics
• Summary page: Key points + action list

Provides a complete Docker containerization deployment solution:

• Includes Python backend image and Node.js frontend image
• nginx as reverse proxy, handling static resources and API forwarding

Deployment steps: After configuring the OpenAI API key, execute docker compose up --build to start the service; Frontend access: localhost:8080; API documentation: localhost:8000/docs.

PawCare+ bridges the information gap between professional veterinary knowledge and ordinary pet owners. Without replacing veterinary diagnosis, it provides timely and personalized health references.

This model reduces unnecessary medical costs, cultivates owners' health awareness and preventive care abilities, and improves pets' quality of life and lifespan in the long run.

1. Organic combination of ML and LLM: ML is responsible for quantitative evaluation and classification, while LLM handles deep understanding and content generation, balancing objectivity and personalization.
2. LangGraph workflow orchestration: Decomposes complex processes into state nodes, making logic easy to maintain and extend, which is valuable for multi-step decision-making AI applications.