On Training Survival Models with Scoring Rules
This work addresses the challenge of improving survival model training for researchers and practitioners in fields like healthcare, offering an incremental advancement by extending scoring rules from evaluation to training.
The authors tackled the problem of training survival analysis models by adapting scoring rules for model training rather than just evaluation, establishing a model-agnostic framework that learns event time distributions parametrically or non-parametrically. The result showed that this approach yields competitive predictive performance with established time-to-event models on synthetic and real-world data.
Scoring rules are an established way of comparing predictive performances across model classes. In the context of survival analysis, they require adaptation in order to accommodate censoring. This work investigates using scoring rules for model training rather than evaluation. Doing so, we establish a general framework for training survival models that is model agnostic and can learn event time distributions parametrically or non-parametrically. In addition, our framework is not restricted to any specific scoring rule. While we focus on neural network-based implementations, we also provide proof-of-concept implementations using gradient boosting, generalized additive models, and trees. Empirical comparisons on synthetic and real-world data indicate that scoring rules can be successfully incorporated into model training and yield competitive predictive performance with established time-to-event models.