Survival Function Matching for Calibrated Time-to-Event Predictions
This work addresses the critical need for calibrated probabilistic predictions in time-to-event models, which is important for risk assessment across diverse applications, representing an incremental improvement over existing methods.
The paper tackles the problem of model calibration in time-to-event predictions, which is under-explored compared to preserving pairwise ordering, by introducing a survival function estimator based on neural networks without explicit distributional assumptions. The result shows that the proposed model outperforms existing approaches in terms of calibration and concentration of time-to-event distributions, as demonstrated in extensive experiments.
Models for predicting the time of a future event are crucial for risk assessment, across a diverse range of applications. Existing time-to-event (survival) models have focused primarily on preserving pairwise ordering of estimated event times, or relative risk. Model calibration is relatively under explored, despite its critical importance in time-to-event applications. We present a survival function estimator for probabilistic predictions in time-to-event models, based on a neural network model for draws from the distribution of event times, without explicit assumptions on the form of the distribution. This is done like in adversarial learning, but we achieve learning without a discriminator or adversarial objective. The proposed estimator can be used in practice as a means of estimating and comparing conditional survival distributions, while accounting for the predictive uncertainty of probabilistic models. Extensive experiments show that the proposed model outperforms existing approaches, trained both with and without adversarial learning, in terms of both calibration and concentration of time-to-event distributions.