Haiku model integrates spatial biology, histology, and clinical data

Researchers have developed Haiku, a novel tri-modal contrastive learning model designed to integrate molecular, morphological, and clinical data from biomedical research. Trained on a large dataset of spatial proteomics and histology images from over 3,000 patients, Haiku creates a shared embedding space for these diverse data types. This enables advanced cross-modal retrieval, improves downstream classification and prediction tasks, and allows for zero-shot biomarker inference, outperforming existing unimodal approaches. AI

影响 Demonstrates a new framework for integrating complex biological data, potentially accelerating discovery in translational medicine.

排序理由 This is a research paper detailing a new model and its performance on specific biomedical tasks. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CV 阅读 →

• arXiv
• breast cancer
• Haiku
• lung cancer

• 论文
• 其他

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →