A1463
Title: Towards scalable survival analysis: Efficient clustering via k-means and log-rank
Authors: Nora Villanueva - University of Vigo (Spain) [presenting]
Marta Sestelo - University of Vigo (Spain)
Luis Machado - University of Minho (Portugal)
Abstract: Survival analysis provides essential tools for studying time-to-event data, with the comparison of survival curves across groups being one of the main objectives. Traditional clustering approaches often rely on bootstrap-based procedures to approximate the null hypothesis distribution. While effective, they impose heavy computational demands and limit scalability in large datasets. The aim is to present a novel method that leverages k-means and the log-rank test to efficiently identify and cluster survival curves. By eliminating the need for intensive resampling, the approach substantially reduces computation time while preserving statistical validity. Through simulation studies, the proposed method is demonstrated to achieve performance comparable to bootstrap-based clustering techniques, while offering a significant gain in efficiency. These findings highlight that the proposed method offers a practical and scalable alternative for the analysis of multiple survival curves.
