Modeling and Predicting Age-at-Onset Trajectories in Genetic Frontotemporal Dementia Using Survival Methods

Ferrández Vicente, José Manuel; Bouzigues, Arabella; Rosen, Howard J.; Höglinger, Günter U.; Forsberg, Leah K.; Heuer, Hilary W.; Wagemann, Olivia; Boxer, Adam L.; Boeve, Bradley F.; Ferry-Bolder, Eve; Val-Calvo, Mikel; Arco, Juan E.; Ramos, E. Marisa; Marth, Lena; Wlasich, Elisabeth; Levin, Johannes; Stockbauer, Anna; Martínez-Murcia, Francisco J.; Ramírez, Javier; Górriz, Juan M.; Foster, Phoebe H.; Schönecker, Sonja; Rohrer, Jonathan D.; Chisman-Russell, Lucy; Adeli, Hojjat; Prix, Catharina; Bernhardt, Alexander Maximilian

doi:10.1007/978-3-032-27314-7_5

Contribution to a conference proceedings/Contribution to a book

DZNE-2026-00658

Modeling and Predicting Age-at-Onset Trajectories in Genetic Frontotemporal Dementia Using Survival Methods

Ferrández Vicente, J. M. (Editor) ; Val-Calvo, M. (Editor) ; Adeli, H. (Editor) ; Arco, J. E. ; Marth, L.DZNE* ; Schönecker, S. ; Martínez-Murcia, F. J. ; Prix, C.Extern* ; Wagemann, O.DZNE* ; Wlasich, E. ; Bernhardt, A. M.Extern* ; Stockbauer, A.Extern* ; Bouzigues, A. ; Chisman-Russell, L. ; Ferry-Bolder, E. ; Foster, P. H. ; Höglinger, G. U.DZNE* ; Rohrer, J. D. ; Boeve, B. F. ; Boxer, A. L. ; Rosen, H. J. ; Heuer, H. W. ; Forsberg, L. K. ; Ramos, E. M. ; Ramírez, J. ; Górriz, J. M. ; Levin, J. (Last author)DZNE*

2026
Springer Nature Switzerland Cham
ISBN: 978-3-032-27313-0 (print), 978-3-032-27314-7 (electronic)

Artificial Intelligence for Neuroscience, Mental Health, and Neurodegenerative Disorders / Ferrández Vicente, José Manuel (Editor) [https://orcid.org/0000-0002-4613-6101] ; Cham : Springer Nature Switzerland, 2026, Chapter 5 ; ISSN: 0302-9743=1611-3349 ; ISBN: 978-3-032-27313-0=978-3-032-27314-7 ; doi:10.1007/978-3-032-27314-7
11th International Work-Conference on the Interplay Between Natural and Artificial Computation, IWINAC 2026, Canary Islands, Spain, 26 May 2026 - 29 May 2026 Cham : Springer Nature Switzerland, Lecture Notes in Computer Science 16574, 45 - 54 (2026) [10.1007/978-3-032-27314-7_5]

This record in other databases:

Please use a persistent id in citations: doi:10.1007/978-3-032-27314-7_5

Abstract: Accurately predicting age at symptom onset in genetic frontotemporal dementia (FTD) remains challenging due to substantial inter-individual variability, even among carriers of distinct pathogenic variants within the same gene. Time-to-event models provide a natural framework to address this problem while accounting for right censoring in presymptomatic individuals. In this study, we analyze age-at-onset prediction in a multicenter cohort of individuals at genetic risk for FTD using a hierarchical survival modeling strategy. We first develop incremental Cox proportional hazards models to systematically evaluate the progressive predictive value of different covariate blocks. We then compare the best-performing Cox model with more flexible non-linear approaches, including DeepSurv and Random Survival Forests. Models are evaluated using strictly out-of-sample five-fold cross-validation, with performance assessed via the concordance index and the integrated Brier score. Our results show that genetic group and progenitor age at onset are the primary drivers of predictive performance, yielding clearly distinct symptom free survival trajectories across risk groups. Non-linear models do not substantially outperform the optimized Cox model, despite their increased flexibility. These findings indicate that Cox models capture most of the prognostic information available in this setting, providing a robust framework for modeling disease trajectories in genetic FTD.

Contributing Institute(s):

Research Program(s):