Blank Bio releases Ladon, an RNA foundation model for predicting patient outcomes

michaelrzhang · x · 2026-11-13

Blank Bio has released Ladon, a transcriptomic foundation model trained on millions of bulk RNA-seq samples, aimed at predicting patient prognosis and treatment response.

Key thesis: AI for drug discovery is booming, but the real bottleneck to deploying drugs is clinical trials. Two patients with the same diagnosis and treatment can have very different outcomes; resolving this heterogeneity is how medicine gets faster.
On held-out external validation cohorts, Ladon outperforms baseline gene-expression and foundation-model methods in predicting prognosis and response.
The post also demonstrates that Ladon finds robust predictive biomarkers from external data, understands isoform usage across tissues and sequencing protocols, and recovers cell-composition signal despite cross-assay batch effects.
The long-term vision is "clinical trial abundance": using predictive models from target indication selection to enrollment criteria so fewer trials fail for avoidable reasons.

Original post →

Blank Bio releases Ladon, an RNA foundation model for predicting patient outcomes

More from Research