KAPLAN: Kolmogorov-Arnold Prognostic Learnable Activation Networks for Survival Analysis
The article introduces KAPLAN-HR, a new model for survival analysis that utilizes Kolmogorov-Arnold Networks. This model aims to improve the estimation of conditional hazards by automatically capturing interactions and time-varying effects without manual specification. Evaluations show that KAPLAN-HR performs comparably or better than existing statistical and deep learning methods on clinical datasets.
- ▪KAPLAN-HR is a B-spline Kolmogorov-Arnold Network designed for nonparametric estimation in survival analysis.
- ▪The model can recover a generalized additive model while deeper architectures can capture complex interactions.
- ▪KAPLAN-HR has been evaluated on six clinical benchmark datasets, demonstrating superior predictive performance.
Opening excerpt (first ~120 words) tap to expand
Statistics > Machine Learning arXiv:2605.23082 (stat) [Submitted on 21 May 2026] Title:KAPLAN: Kolmogorov-Arnold Prognostic Learnable Activation Networks for Survival Analysis Authors:Stelios Boulitsakis Logothetis, Angela Wood, Pietro Li ò View a PDF of the paper titled KAPLAN: Kolmogorov-Arnold Prognostic Learnable Activation Networks for Survival Analysis, by Stelios Boulitsakis Logothetis and 2 other authors View PDF HTML (experimental) Abstract:Survival analysis aims to model how covariates and time jointly shape the time-to-event distribution under right censoring. Classical methods such as the Cox model and generalised additive models (GAMs) require interactions and time-varying effects to be manually specified, which is increasingly impractical on rich clinical datasets.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
