Skill-Augmented AI Agents for Medical Research Analysis: An Exploratory Multi-Model Human Evaluation in an NSCLC Transcriptomic Biomarker Task

This study evaluates the effectiveness of skill-augmented AI agents in generating high-quality transcriptomic research outputs for non-small cell lung cancer biomarker analysis, utilizing six model backbones and the OpenClaw implementation. Results indicated that skill-augmented outputs had a mean expert-rated quality of 5.50 compared to 5.11 for native AI, although the differences were not statistically significant. The findings suggest that while skill augmentation may enhance output quality, further research is needed to validate these results and assess reliability in larger studies, highlighting the importance of integrating skills into AI systems for biomedical applications.