Gast ::
| Hauptseite > Publikationsdatenbank > Predicting individual hand grip strength - a multimodal, confound-free machine learning approach |
| Hauptseite > Publikationsdatenbank > Predicting individual hand grip strength - a multimodal, confound-free machine learning approach |
| Conference Presentation (Invited) | FZJ-2024-07537 |
; ; ; ; ;
2024
Abstract: Hand grip strength (HGS) not only reflects overall strength, but is also closely related to physical disability, cognitive decline and mortality [2,8,5] . Beyond, HGS is a cost-efficient and reliable measure in clinical practice. Despite its ubiquity, the neural mechanisms governing HGS remain unclear. To reveil neural underpinnings driving out-of-sample prediction of HGS, we investigated 9 neuroimaging-derived feature categories in the UK Biobank [3] (N = 22554-33136) in combination with 7 algorithms and ensembles thereof under 5 confound-removal scenarios. Additionally we trained models on sex-split populations to rule out non-linear sex-influences. Only such confound-free models allow for the aimed neural interpretation of predictions. Under the most stringent confounder control, inputting grey matter volume (GMV), fALFF and a collection of 6 white matter microstructural characteristics to the XGBoost algorithm yielded significantly best predictions. Interpretative SHAP analyses reveiled that GMV in the anterior globus pallidus and microstructural characteristics of sensory input bundles to the thalamus and thalamo-cortical tracts were driving out of sample prediction of HGS. This not only informs us about the single-subject-level neural underpinnings of motor behaviour. Being in line with insights from functional neuroanatomy, our results also bridge a gap between the micro- and macrolevel neuroscientific understanding of motor behaviour.
|
The record appears in these collections: |