The COSMIC Knowledgebase, the world’s leading expert-curated resource for somatic mutations in cancer, combines rigorous manual review with scalable computational approaches to keep pace with rapidly expanding scientific literature. In this webinar, we explore how artificial intelligence (AI) is integrated into the COSMIC curation workflow to support – rather than replace – expert decision-making.

 
Using a large, long-established curated dataset, COSMIC Knowledgebase has trained and evaluated AI models to pre-screen PubMed search results and identify papers likely to contain curatable mutation data. This enables efficient triage of hundreds of publications weekly and reduces manual workload while maintaining high data quality.
 
We also showcase an AI-assisted workflow currently being developed by the COSMIC Actionability team to identify potentially relevant clinical trials from ClinicalTrials.gov. AI helps detect biomarkers in the appropriate clinical context and supports automated filtering, while final inclusion decisions are made by expert manual review.
 
In this webinar, we will explore the development, performance and application of COSMIC’s AI curation pipeline, including its limitations, ongoing refinement and how human expertise remains essential to validation, prioritization and interpretation.
 
What you will learn:

• - How COSMIC integrates AI into a human-led curation workflow to pre-screen and prioritize scientific literature

• - How AI models are trained, evaluated and applied to identify papers likely to contain curatable mutation data

• - The strengths, limitations and performance of current AI approaches, and where expert curators remain essential

• Future directions for scaling high-quality, AI-assisted cancer mutation curation