An interpretable AI explains why immune cells fail to reach some tumours
Most spatial-biology tools show where cells sit. This one tries to say why.
作者：The desk, News · · 1 分钟阅读
标签: biology · cancer · immunology · nature
这篇报道还没有中文（简体）版本，因此你正在阅读英文。
- 这一则
- Most spatial-biology tools show where cells sit.
- 为何重要
- This one tries to say why.
- 已确认的部分
- The photograph is generated for this story.
对应已点名的来源。未标为已核实。 Nature Biomedical Engineering: Interpretable AI model reveals spatial codes of cell recruitment in tissues
The news
A paper published in Nature Biomedical Engineering on October 13 introduces SPACER, a fully interpretable machine-learning framework that decodes gene-level recruitment rules from spatially resolved transcriptomics data. Applied to 37 datasets spanning several human tumour types and mouse heart, it achieved 4- to 21-fold improvements over existing methods in recovering biologically meaningful recruitment signals, the authors report.
Why it matters
Knowing why T cells fail to engage a tumour could help predict who responds to immunotherapy. An interpretable model gives biologists rules they can test, not just a score.
The details
SPACER found that mucins expressed by tumours are associated with physical barriers to T cell engagement, which predicts resistance to immune checkpoint inhibitors and patient prognosis. It identified local antigen presentation as the dominant correlate of T cell engagement across cancer types.
It also pointed to CD4-positive T cells as the dominant effectors in severe myocarditis caused by checkpoint inhibitors.
What we could not check
- The full paper is behind a paywall; we relied on the abstract.
- The findings are associations that need experimental confirmation.
来源
The photograph is generated for this story. It does not show the people named in it.