{"data":{"kind":"file","path":"README.md","version_id":"enkggly7tggclnoyncx9n7d7","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":2325,"modified_at":"2026-01-29T12:03:30.733000","content_hash":"177b924520e2cebe3a9bf830bb6105c0f8834979badd7047198dbf9fe9eb47f2"},"entries":[],"content":"# OpenMed HealthFact Environment\n\nHealth claim fact-checking environment for RL fine-tuning using the PubHealth dataset.\n\n## Task Description\n\nGiven a health-related claim, classify its veracity:\n\n| Class | Label | Description |\n|-------|-------|-------------|\n| A | TRUE | Claim is accurate and supported by evidence |\n| B | FALSE | Claim is false or misleading |\n| C | UNPROVEN | Claim cannot be verified with available evidence |\n| D | MIXTURE | Claim contains both true and false elements |\n\n## Dataset\n\n- **Source**: PubHealth (Kotonya & Toni, EMNLP 2020)\n- **Processed**: [OpenMed/PubHealth-Processed](https://huggingface.co/datasets/OpenMed/PubHealth-Processed)\n- **Size**: 12,253 claims\n  - Train: 9,805\n  - Validation: 1,215\n  - Test: 1,233\n- **Topics**: Biomedical subjects, healthcare policy, public health\n\n## Reward Structure\n\n| Component | Weight | Description |\n|-----------|--------|-------------|\n| Accuracy | 80% | Exact match on fact-check classification (A-D) |\n| Thinking | 15% | Quality of reasoning in `<think>` tags |\n| Format | 5% | Proper `\\boxed{}` answer format |\n\n## Example\n\n**Input:**\n```\nClaim: \"Vitamin C prevents the common cold.\"\n\nRelated topics: vitamins, cold, prevention\n```\n\n**Expected Output:**\n```\n<think>\nThe claim states that vitamin C prevents the common cold. While vitamin C has been\nstudied extensively for cold prevention, research shows mixed results. Meta-analyses\nindicate that regular vitamin C supplementation may slightly reduce cold duration in\nsome populations, but does not prevent colds in the general population. The claim\noverstates the preventive effect - it's partially true (may reduce duration) but\nfalse regarding prevention.\n</think>\n\\boxed{D}\n```\n\n## Usage\n\n```python\nfrom OpenMed_HealthFact import load_environment\n\nenv = load_environment()\n```\n\n## Citation\n\n```bibtex\n@inproceedings{kotonya-toni-2020-explainable,\n    title = \"Explainable Automated Fact-Checking for Public Health Claims\",\n    author = \"Kotonya, Neema  and Toni, Francesca\",\n    booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)\",\n    year = \"2020\",\n    url = \"https://aclanthology.org/2020.emnlp-main.623\",\n    doi = \"10.18653/v1/2020.emnlp-main.623\",\n    pages = \"7740--7754\",\n}\n```\n\n## License\n\nMIT (original PubHealth dataset)\n","encoding":"utf-8","truncated":false,"total_bytes":2325},"status":null}