SurvICL: a Foundation Model for in-context Survival Analysis
In the authors' words
Tabular foundation models (TFMs) now outperform gradient-boosted trees in classification and regression. Yet in survival analysis, which predicts when an event occurs from censored data, often on small clinical cohorts, we find that no existing TFM beats the half-century-old Cox model. We address this issue by introducing SurvICL, a TFM that predicts survival curves in-context. SurvICL is pretrained on purely synthetic datasets generated from a novel flexible proportional-odds prior that outperforms other prior types in our ablation study. Unlike most other survival models, SurvICL is trained with a plain quantile regression loss, which is possible because its synthetic pre-training data generator can provide uncensored event times. Additionally, we introduce the largest survival benchmark to date spanning 103 datasets. On the 63 datasets held out from all design decisions, SurvICL outperforms 20 tree-based, deep, and foundation-model baselines without any tuning: it wins 82.9% of pairwise comparisons in integrated Brier score, against 71.3% for the runner-up, the Cox model, and is the first model to rank above Cox on small datasets.
In plain words AI summary
Survival analysis predicts when an event will happen from incomplete data, often in small medical studies. The authors created SurvICL, a foundation model that predicts survival curves in-context. SurvICL was pretrained only on synthetic data from a new flexible prior that beat other priors in tests. On 63 test datasets, SurvICL beat 20 other methods without any tuning, winning 82.9% of comparisons. It is the first model to rank above the classic Cox model on small datasets.
Written by Kimi K2.5 using only the abstract above. Jev checked that every sentence is supported by the abstract sentences it cites, and removed any that were not. Point at a sentence to highlight its source.
Appeared: Friday, October 16. arXiv. Preprint, not yet peer-reviewed.