Overview

This dataset contains 2,000 LLM-generated pet health symptoms text samples covering 5 common pet health condition categories, designed to train ML models for automated pet health classification. Each entry is labeled with: - Pet health condition (1 of 5 distinct classes)
- Record type (Owner Observation or Clinical Notes)

Owner observations are expressed in everyday language (e.g., "My cat scratches constantly"), whereas clinical notes contain veterinary terminology (e.g., "Pruritus with alopecia").

Features

• text: One concise sentence describing the pet health symptoms
• condition: Skin Irritations, Digestive Issues, Parasites, Ear Infections, Mobility Problems
• record_type: Owner Observation, Clinical Notes

Classification Types

• Binary Classification (record_type)
• Multi-class Classification (condition)
• Multi-task Classification (condition and record_type)

Potential Use Cases

• Pet health chatbots for owners to assess symptoms
• Automated triage systems for veterinary clinics
• Educational tools for vet student training
• Insurance claim processing for analyzing unstructured pet health records
• EHR integration (structuring historical clinical notes)

Limitations

• Synthetic data: Simulated but not real clinical records
• Limited scope: Covers only 5 common conditions
• Species bias: Primarily cats/dogs (few exotic pet examples)
• Language simplicity: Lacks complex medical edge cases
• Demographic gaps: No age/breed metadata for analysis

Provenance

• Synthetic data generated by Gemini 2.5 Pro Experimental, with prompts stating output format, condition classes, record_type classes, number of samples required, and "maximize your creativity, generate unique data, avoid duplicate data"
• Logo image generated by Mistral AI Le Chat (Black Forest Labs Flux Ultra) with the prompt "Silhouettes of vet and pets, overlapping circles with health issue symbols."

Pet Health Symptoms Dataset

  Overview

This dataset contains 2,000 LLM-generated pet health symptoms text samples covering 5 common pet health condition categories, designed to train ML models for automated pet health classification. Each entry is labeled with:

Pet health condition (1 of 5 distinct classes)
Record type (Owner Observation or Clinical Notes)

Owner observations are expressed in everyday language (e.g., "My cat scratches constantly"), whereas clinical… See the full description on the dataset page: https://huggingface.co/datasets/karenwky/pet-health-symptoms-dataset.