Su, W., Liu, K., Yin, G., Huang, J., & Zhao, X. (2024). Deep Nonparametric Inference for Conditional Hazard Function. arXiv preprint arXiv:2410.18021.
This paper aims to develop a flexible and robust method for estimating conditional hazard functions in survival analysis, addressing limitations of traditional methods that rely on strong assumptions about the underlying data distribution.
The authors propose a novel approach using deep neural networks (DNNs) to approximate the logarithm of the conditional hazard function directly. They establish the nonasymptotic error bound and asymptotic properties of the DNN estimator, proving its consistency and functional asymptotic normality. Based on this framework, they develop one-sample and two-sample tests for comparing conditional hazard functions and a goodness-of-fit test for evaluating model adequacy.
The DNN-based approach provides a powerful and flexible framework for nonparametric inference of conditional hazard functions in survival analysis. It offers advantages over traditional methods by relaxing restrictive assumptions and effectively capturing complex relationships between covariates and survival outcomes.
This research significantly contributes to the field of survival analysis by introducing a novel DNN-based methodology for nonparametric inference. It offers a practical and robust alternative to traditional methods, potentially leading to more accurate and reliable estimations and inferences in various applications involving time-to-event data.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Wen Su, Kin-... at arxiv.org 10-24-2024
https://arxiv.org/pdf/2410.18021.pdfDeeper Inquiries