Reproduce “Learning Scientific Trajectories to Predict Future Biomedical Discoveries” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chia-Hsuan Chang, Brian Ondov, Jiang Lv, Lan Jiang, Yunmai Wang, Haerin Song, Xueqing Peng, Qiaozhu Mei and 2 more
ForeSCITE generates biomedical hypotheses by learning an early-to-late direction in topic-specific conclusion embeddings, shifting existing conclusions along it, and decoding them into text. In retrospective tests, it exceeded evaluated prompt-based and retrieval-augmented baselines in semantic coverage, reaching about 90% and 77% of human post-cutoff coverage at α = 0.8. This measures embedding-space coverage, not correctness or prospective validity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.