CoxSE : Exploring the potential of self-explaining neural networks with Cox proportional hazards model for survival analysis
Alabdallah, Abdallah ; Hamed, Omar ; Ohlsson, Mattias LU
; Rögnvaldsson, Thorsteinn
and Pashami, Sepideh
(2026)
In Knowledge-Based Systems
333.
- Abstract
The Cox Proportional Hazards (CPH) model has long been the preferred survival model for its explainability. However, to increase its predictive power beyond its linear log-risk, it was extended to utilize deep neural networks, sacrificing its explainability. In this work, we explore the potential of self-explaining neural networks (SENN) for survival analysis. We propose a new locally explainable Cox proportional hazards model, named CoxSE, by estimating a locally-linear log-hazard function using the SENN. We also propose a modification to the Neural additive (NAM) model, hybrid with SENN, named CoxSENAM, which enables the control of the stability and consistency of the generated explanations. Several experiments using synthetic and... (More)
The Cox Proportional Hazards (CPH) model has long been the preferred survival model for its explainability. However, to increase its predictive power beyond its linear log-risk, it was extended to utilize deep neural networks, sacrificing its explainability. In this work, we explore the potential of self-explaining neural networks (SENN) for survival analysis. We propose a new locally explainable Cox proportional hazards model, named CoxSE, by estimating a locally-linear log-hazard function using the SENN. We also propose a modification to the Neural additive (NAM) model, hybrid with SENN, named CoxSENAM, which enables the control of the stability and consistency of the generated explanations. Several experiments using synthetic and real datasets are presented, benchmarking CoxSE and CoxSENAM against a NAM-based model, a DeepSurv model explained with SHAP, and a linear CPH model. The results show that, unlike the NAM-based model, the SENN-based model can provide more stable and consistent explanations while maintaining the predictive power of the black-box model. The results also show that, due to their structural design, NAM-based models demonstrate better robustness to non-informative features. Among the models, the hybrid model exhibits the best robustness. Full implementation is available on GitHub.
(Less)
- author
- Alabdallah, Abdallah
; Hamed, Omar
; Ohlsson, Mattias
LU
; Rögnvaldsson, Thorsteinn
and Pashami, Sepideh
- organization
-
- Computational Science for Health and Environment (research group)
- Department of Earth and Environmental Sciences (MGeo)
- LU Profile Area: Natural and Artificial Cognition
- Centre for Environmental and Climate Science (CEC)
- eSSENCE: The e-Science Collaboration
- Artificial Intelligence in CardioThoracic Sciences (AICTS) (research group)
- publishing date
- 2026-01-30
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Cox proportional hazards, Interpretability, Neural additive models, Self-explaining neural networks, Survival analysis, XAI
- in
- Knowledge-Based Systems
- volume
- 333
- article number
- 114996
- publisher
- Elsevier
- external identifiers
-
- scopus:105024190200
- ISSN
- 0950-7051
- DOI
- 10.1016/j.knosys.2025.114996
- language
- English
- LU publication?
- yes
- id
- f2c2cf82-0871-4dc2-a580-85b0b4081bba
- date added to LUP
- 2026-03-09 15:08:42
- date last changed
- 2026-03-09 15:08:49
@article{f2c2cf82-0871-4dc2-a580-85b0b4081bba,
abstract = {{<p>The Cox Proportional Hazards (CPH) model has long been the preferred survival model for its explainability. However, to increase its predictive power beyond its linear log-risk, it was extended to utilize deep neural networks, sacrificing its explainability. In this work, we explore the potential of self-explaining neural networks (SENN) for survival analysis. We propose a new locally explainable Cox proportional hazards model, named CoxSE, by estimating a locally-linear log-hazard function using the SENN. We also propose a modification to the Neural additive (NAM) model, hybrid with SENN, named CoxSENAM, which enables the control of the stability and consistency of the generated explanations. Several experiments using synthetic and real datasets are presented, benchmarking CoxSE and CoxSENAM against a NAM-based model, a DeepSurv model explained with SHAP, and a linear CPH model. The results show that, unlike the NAM-based model, the SENN-based model can provide more stable and consistent explanations while maintaining the predictive power of the black-box model. The results also show that, due to their structural design, NAM-based models demonstrate better robustness to non-informative features. Among the models, the hybrid model exhibits the best robustness. Full implementation is available on GitHub.</p>}},
author = {{Alabdallah, Abdallah and Hamed, Omar and Ohlsson, Mattias and Rögnvaldsson, Thorsteinn and Pashami, Sepideh}},
issn = {{0950-7051}},
keywords = {{Cox proportional hazards; Interpretability; Neural additive models; Self-explaining neural networks; Survival analysis; XAI}},
language = {{eng}},
month = {{01}},
publisher = {{Elsevier}},
series = {{Knowledge-Based Systems}},
title = {{CoxSE : Exploring the potential of self-explaining neural networks with Cox proportional hazards model for survival analysis}},
url = {{http://dx.doi.org/10.1016/j.knosys.2025.114996}},
doi = {{10.1016/j.knosys.2025.114996}},
volume = {{333}},
year = {{2026}},
}