Lise Gamboa - Interpretation measures in Random Survival Forests (RSF) with longitudinal predictors

par Lise Gamboa (Université de Bordeaux, Inserm Bordeaux Population Health U1219, BIOSTAT)

jeudi 18 déc. 2025, 13:00 → 14:00 Europe/Paris

Module 1.3 (ISPED)

Speaker: Lise Gamboa from BPH
Title: Interpretation measures in Random Survival Forests (RSF) with longitudinal predictors

Abstract: 

In many biomedical studies, time-to-event outcomes must be predicted from diverse predictors, including numerical, categorical, and longitudinal variables measured repeatedly over time. Random Survival Forests (RSF) are increasingly used in this setting because they provide strong predictive performance, handle complex relationships, and do not rely on assumptions such as proportional hazards ratio. However, their internal mechanisms remain difficult to interpret, which limits their use in clinical contexts where interpretability is essential. Existing interpretability tools such as variable importance, partial dependence plots, or SHAP values offer useful insights but are not yet adapted to longitudinal predictors, even though these dynamic markers have critical information in health research.

This work addresses this limitation by exploring how current interpretability methods can be extended to RSF models that incorporate longitudinal data. To do so, we rely on the DynForest package, which constructs random forests specifically designed to integrate longitudinal markers into the construction of the forest. The aim is to improve the interpretability of RSF in settings where repeated measurements play a central role.

Calendar subscription link for the complete seminar series:
https://indico.math.cnrs.fr/category/711/events.ics

Program of the Biostatistics seminars:
https://indico.math.cnrs.fr/category/711/

Subscribe to the seminar mailing list:
https://diff.u-bordeaux.fr/sympa/subscribe/seminaire.biostat.bph

Former e-seminars on our YouTube channel (mostly in French): https://www.youtube.com/channel/UCURp-hEQL7k23UzGfqgEurA/videos

 

Biostatistics seminar series from the Department of Public Health from the University of Bordeaux and the Bordeaux Population Health UMR 1219 research center