Learning Survival Models with Right-Censored Reporting Delays
Organizations: The Graduate University for Advanced Studies · Tokio Marine Holdings, Inc. · The Institute of Statistical Mathematics · RIKEN
Abstract
Survival analysis provides statistical methods to model the time until an event occurs. Reporting delays arise when event times are not observed at their occurrence but are only revealed upon reporting. This issue is particularly critical for timely risk evaluation when the observation window is short due to administrative censoring. In this study, we incorporate right-censored reporting delays by jointly modeling parametric hazards for the event and reporting processes. We then construct a consistent estimator for the model parameters and develop a Monte Carlo expectation-maximization algorithm to compute it. To address the challenges posed by administrative censoring, we leverage these findings and propose a transfer-learning procedure. Experimental results demonstrate that our method improves the accuracy of timely risk evaluation under administrative censoring.
Figures & tables
| Dataset | Method | ||||
| Standard Direct | |||||
| Standard | |||||
| Toy1 | Ours Direct | ||||
| Ours MCEM | |||||
| Ours | |||||
| Standard Direct |
Appendix figures & tables2 assets
Supplementary material from the paper’s appendix.
Appendix
| Notation | Description |
| the combined pseudo-complete dataset corresponding to the earlier process | |
| the combined pseudo-complete dataset corresponding to the later process | |
| the number of observations |
| Parameter | Toy1 | Toy2 | |||||||
| Bias | RMSE | MeanSE | Coverage | Bias | RMSE | MeanSE | Coverage | ||
| 0.014 | 0.178 | 0.166 | 0.924 | 0.001 | 0.100 | 0.094 | 0.942 | ||
| -0.111 | 0.272 | 0.239 | 0.936 | -0.050 | 0.182 | 0.169 | 0.944 | ||
| -0.092 | 0.259 | 0.238 | 0.938 | -0.034 | 0.172 | 0.164 | 0.942 | ||
| -0.029 | 0.259 | 0.238 | 0.926 | -0.013 | 0.143 | 0.144 | 0.958 | ||
| 0.051 | 0.236 | 0.218 | 0.928 | 0.005 | 0.160 | 0.151 | 0.932 | ||
Survival Analysis
Survival analysis models when events such as disease progression, failure, or death occur, accounting for incomplete follow-up and censoring.
See all 112 papers