An agent generated a cancer hypothesis, and the lab confirmed it
- Source
- x.com
- Date
AI’s biggest role in science may not be answering questions. It may be helping scientists find which questions are worth asking. At @ProvSwedish, AutoDiscovery surfaced an unexpected immune signal in a heavily studied cancer dataset—and follow-up research confirmed it. 🧵
@ProvSwedish The signal involved invasive lobular carcinoma (ILC), a subtype affecting ~48K Americans each year. ILC has long been considered “immune cold,” with relatively low immune activity. AutoDiscovery found more activity than expected.
@ProvSwedish Providence Swedish researchers ran AutoDiscovery over The Cancer Genome Atlas (TCGA), a resource built from donated cancer patient samples. AutoDiscovery generated + tested hypotheses across the data, searching for results that challenged expectations.
@ProvSwedish One was the ILC immune signal. The researchers confirmed the same pattern in separate patient data. Lab analysis subsequently verified it, finding T-cells around ILC tumors.
Context
Ai2's AutoDiscovery, a system that uses language models to generate and test hypotheses on existing datasets, helped researchers identify an unexpected immune pattern in lobular breast cancer, but expert guidance was consequential. The researchers' preprint, which is not yet peer reviewed, reports that an unguided run produced no clinically meaningful novel findings. A run seeded with minimal input from an oncologist produced several interesting hypotheses, including a stronger immune signature in this cancer type.
The authors report that they checked the pattern in independent patient cohorts and confirmed supporting immune-cell evidence in tumor tissue. That sequence connects an AI-generated question to biological checking. It gives researchers a reason to investigate immunotherapy in this cancer subtype; it does not establish that such treatment benefits patients. Reported validation.
Checking sign-in…
Loading comments…