Drtamar/veterinary-fewshot-diagnostics

🐾 Veterinary Multimodal Diagnostic System

AI-powered veterinary diagnostic system using Few-Shot Learning with Prototypical Networks for multi-label disease classification in cats and dogs.

🎯 Features

• —Few-Shot Learning - Works with only 5 examples per disease
• —11 Disease Detection - Babesiosis, Ehrlichiosis, CKD, AKD, and more
• —Multimodal Integration - Blood, urine, vitals, biochemistry, demographics
• —Real-time Predictions - Instant diagnostic support
• —Interactive UI - User-friendly Streamlit interface

🚀 Quick Start

1. 1.Upload 5 CSV files
2. 2.Select an encounter
3. 3.Review patient data
4. 4.Generate prediction

🏥 Detectable Conditions

1. 1.Healthy
2. 2.Tick-Borne Disease
3. 3.Babesiosis
4. 4.Ehrlichiosis
5. 5.Filariasis
6. 6.Urinary Tract Infections
7. 7.Chronic Kidney Disease (CKD)
8. 8.Acute Kidney Disease (AKD)
9. 9.Hepatozoon Infection
10. 10.IMHA
11. 11.Pneumonia

🧠 Architecture

• —Model: Prototypical Network
• —Embedding Dim: 256
• —Support Set: 65 examples (5-shot)
• —Features: 207
• —Threshold: 30%

⚠️ Disclaimer

Research prototype - NOT for clinical use. Always consult qualified veterinarians.

👨‍💻 Author

Udavith-Reshanjana

📄 License

MIT