Discovering Prognostic Risk Strata and Heterogeneous Progression Routes from Cross-Sectional Survival Data

GOTSurv infers patient-specific progression routes from cross-sectional survival data, stratifying TCGA patients into prognostic risk groups and revealing distinct within-stratum trajectories linked to rising risk.

• Why it matters: Longitudinal follow-up is often missing, and cross-sectional analyses can miss heterogeneous individual progression patterns.
• What they did: GOTSurv learns survival-risk rankings and automatically forms prognostic strata without predefined cut-offs, then applies unbalanced optimal transport to estimate feature changes tied to transitions toward higher-risk groups.
• The result: Across ten TCGA cancer cohorts (gene expression n=4,767; whole-slide images n=4,655) with external validation in two independent cohorts, it generally strengthens prognostic stratification versus established models while uncovering individualized progression trajectories within the same risk stratum.

AI summary of the abstract · may contain errors · not medical advice · read the paper.