Poster (After Call)

FZJ-2022-02636

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Brain-age prediction: a systematic comparison of machine learning workflows

More, S. (Corresponding author)FZJ* ; Antonoupolous, G.FZJ* ; Hoffstaedter, F.FZJ* ; Caspers, J. ; Eickhoff, S.FZJ* ; Patil, K.FZJ*

2022

Please use a persistent id in citations: http://hdl.handle.net/2128/31469

Abstract: Prediction of age using anatomical brain MRI, i.e., brain age, is proving valuable in exploring accelerated aging (brain age delta) as a proxy for aging-related diseases and crucial future health outcomes [1]. While various data representations and machine learning (ML) algorithms have been used for brain-age prediction [2,3], the impact of these choices on prediction accuracy remains uncharacterized. Moreover, several methodological challenges remain before a predictive model can be deployed in the real world; (1) robust within-site performance, (2) accurate cross-site prediction and, (3) consistent prediction for the same individual. To fill this gap, we systematically evaluated 70 workflows consisting of ten feature spaces derived from grey matter (GM) images and seven ML algorithms with diverse inductive biases to establish guidelines for designing brain-age prediction workflows.

---

Contributing Institute(s):

1. Gehirn & Verhalten (INM-7)

Research Program(s):

1. 5251 - Multilevel Brain Organization and Variability (POF4-525) (POF4-525)
2. 5254 - Neuroscientific Data Analytics and AI (POF4-525) (POF4-525)

Appears in the scientific report 2022

---

Database coverage:

---