How Researchers Claim Novelty in Biomedical Science: A Taxonomy for Understanding Innovation

Scientific novelty is essential for progress, yet it is often hard to define and measure. In this study, we create NOBL(NOvelty taxonomy for Biomedical Literature), a hierarchical taxonomy for classifying different types of noveltyclaims in biomedical research papers. Developed through an iterative, data-driven, expert-guided, and AI-assistedframework using a domain-stratified corpus of 12,701 PubMed articles, NOBL offers a fine-grained, interpretable schema of innovation types, validated for clarity, consistency, and coverage. We find that 6.47% of articles include at least one explicit novelty claim, with new findings being the most common type. We also identify three distinct innovation patterns across 130 biomedical fields. NOBL provides a systematic approach for researchers, reviewers, and funders to understand the innovation landscape of biomedical research and offers a new way to make scientific novelty more visible, measurable, and useful.