Skip to content

Dataset · curaihealth

Physician-labeled English medical question pairs

curaihealth/medical_questions_pairs

A set of 3,048 similar and dissimilar medical question pairs manually authored and labeled by physicians.

What's inside

Inside are 3,048 English question pairs for a text-classification task. The foundation is 1,524 real patient questions from the HealthTap platform, for each of which Curai physicians manually wrote one semantically similar and one distinct question.

What you can do with it

  • Train models for duplicate detection and semantic similarity of medical queries
  • Fine-tune embeddings for the medical domain in English
  • Test patient question routing systems in telemedicine
Tasks
Text classification
Languages
English
Size
1K–10K rows
Downloads this month
1.8K
Likes
52

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("curaihealth/medical_questions_pairs")