Identifying conflicting claims in biomedical literature is critical for advancing scientific understanding, yet the scarcity of high-quality training data remains a significant challenge. We introduce EvoNLI, an evolutionary algorithm that learns how to transform entailing sentence pairs into challenging contradictions by mutating words until a frozen teacher model confidently flips its prediction, while preserving topical coherence. EvoNLI, applied to PubMed randomized controlled trials (RCTs), generates SciCon—a dataset of premise–hypothesis pairs automatically labeled with 94.4% precision, as verified by domain experts. Fine-tuning large language models on SciCon improves contradiction ROC-AUC consistently across eight biomedical NLI benchmarks. EvoNLI and SciCon are publicly available to support contradiction-aware biomedical search and evidence synthesis, and to advance robust domain-specific contradiction detection.
