Researchers introduce PathLUPI to improve molecular prediction from histology images
A team of researchers presented PathLUPI, a new framework that incorporates transcriptomic privileged information during training to create genome‑anchored embeddings for whole‑slide images (WSIs). At inference, the model relies only on routine pathology slides, yet it captures molecular context that improves prediction of complex biomarkers and patient outcomes.\n\nThe method was tested on 49…
1 source primary source
Key points
- PathLUPI trains genome‑anchored histology embeddings using transcriptomic privileged information, enabling WSI‑only inference.
- Evaluated on 49 tasks, 11,361 cases from 20 cohorts, achieving AUC ≥0.80 on 18 biomarkers and C‑index ≥0.70 on five cancer survival cohorts.
- Embeddings expose genotype‑linked morphological patterns confirmed by high‑resolution spatial transcriptomics.
Read the original at Nature Machine Learning primary sourceOpen source ↗
Topics · follow one to build your own front page
PathLUPIC. JinF. ZhouY. Yu
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Empero Research releases PiVo-1.4M, a 1.4M-parameter text-to-speech model for local use · 1 src
- Unite.AI explains AI latency and throughput metrics for engineers · 1 src
- SemiAnalysis releases GDPx benchmark with Claude Opus 5.5 leading at 12% · 3 src
- Gemma-3-12b fails to detect misinformation when poisoned claims dominate retrieved evidence · 1 src
- Researchers release BILAL-BENCH to test Bengali AI models’ Islamic text fabrication · 1 src
Comments
via GitHub Discussions