MedNLI
by University of Massachusetts / Partners Healthcare · free · Last verified 2026-03-17
MedNLI is a benchmark dataset for Natural Language Inference (NLI) in the clinical domain. Derived from the MIMIC-III database, it contains over 14,000 sentence pairs from clinical notes, each annotated by a clinician as representing entailment, contradiction, or a neutral relationship, enabling the evaluation of clinical text reasoning.
https://physionet.org/content/mednli/ ↗B
B—Above Average
Adoption: BQuality: AFreshness: BCitations: B+Engagement: F
Specifications
- License
- PhysioNet Credentialed Health Data License
- Pricing
- free
- Capabilities
- Clinical Natural Language Inference (NLI), Semantic Textual Similarity Assessment, Contradiction Detection in Clinical Narratives, Training Domain-Specific Language Models, Benchmarking Clinical Text Understanding, Evaluating Factual Consistency in Summaries, Fine-tuning Foundation Models for Healthcare
- Integrations
- [object Object], [object Object], [object Object], [object Object]
- Use Cases
- [object Object], [object Object], [object Object], [object Object]
- API Available
- No
- Tags
- natural-language-inference, clinical-nlp, entailment, mimic, benchmark, healthcare-ai, text-classification, semantic-reasoning, medical-records, llm-evaluation, contradiction-detection
- Added
- 2026-03-17
- Completeness
- 0.8%
Index Score
62.8Adoption
65
Quality
89
Freshness
60
Citations
76
Engagement
0