This title appears in the Scientific Report :
2023
Please use the identifier:
http://dx.doi.org/10.34734/FZJ-2023-03043 in citations.
Predicting Executive Functioning from Brain Networks: Modality Specificity and Age Effects
Predicting Executive Functioning from Brain Networks: Modality Specificity and Age Effects
Healthy aging is associated with structural and functional brain changes, which have beenlinked to executive functioning (EF) decline. Despite a known association between resting-statefunctional connectivity (RSFC) and EF [1], its potential for capturing individual differences inEF performance has b...
Saved in:
Personal Name(s): | Heckner, Marisa (Corresponding author) |
---|---|
Cieslik, Edna / Paas Oliveros, Lya Katarina / Eickhoff, Simon / Patil, Kaustubh / Langner, Robert | |
Contributing Institute: |
Gehirn & Verhalten; INM-7 |
Imprint: |
2023
|
DOI: |
10.34734/FZJ-2023-03043 |
Conference: | Organization for Human Brain Mapping (OHBM), Montreal (Canada), 2023-07-22 - 2023-07-26 |
Document Type: |
Poster |
Research Program: |
JL SMHB - Joint Lab Supercomputing and Modeling for the Human Brain (JL SMHB-2021-2027) Brain Dysfunction and Plasticity |
Link: |
OpenAccess |
Publikationsportal JuSER |
Healthy aging is associated with structural and functional brain changes, which have beenlinked to executive functioning (EF) decline. Despite a known association between resting-statefunctional connectivity (RSFC) and EF [1], its potential for capturing individual differences inEF performance has been questioned [2]. Therefore, we examined to what degree individual EFabilities may be predicted from different brain modalities: RSFC, grey-matter volume (GMV),regional homogeneity (ReHo), and fractional amplitude of low frequency fluctuations (fALFF).This was done in an EF-associated (EFN), a perceptuomotor (PercMot) and a whole-brainnetwork in young and old adults.We meta-analytically defined an EFN [3] and a perceptuomotor (linked to visual,auditory, and motor processing) network [4] and used Power et al.’s [5] graph of putativefunctional areas as a whole-brain control. Imaging and behavioral data of 116 younger (age: 20–40 years, 64 females) and 111 older (age: 60–80 years, 72 females) healthy adults were obtainedfrom the enhanced NKI sample [6]. Targets for prediction comprised performance in highlydemanding (HD) and less demanding (LD) conditions of each of 3 classic EF tasks: Color-WordInterference, Trail Making, and N-Back. Individual z-transformed performance scores were thenpredicted from the characteristics of each network’s edges (RSFC) or nodes (GMV, ReHo,fALFF) using partial least-squares regression with 100 repetitions of a 10-fold cross-validationscheme. Prediction accuracy as indicated by the root mean square error (RMSE) for the 100repetitions was then submitted to a 2(age group) × 3(network) × 2(task demand level) ×4(modalities) mixed-measures ANOVA.Overall, prediction accuracy (i.e., RMSE) was rather low to moderate. Similar toprevious studies [2;7], prediction accuracy was better for older [M = .784] (vs. younger [M =.794]) adults, indicating that brain–behavior association strength increases with advancing age.This might be due to overall age-related neural decline such as atrophy or white-matterdegeneration, changes that have been associated with network integrity [8] and reorganization aswell as EF decline. Interestingly, prediction accuracy for younger adults was better whenpredicting HD [M = .757] (vs. LD [M = .831]) conditions, while for older adults, predictionaccuracy was better when predicting LD [M = .771] (vs. HD [M = .797]) conditions. Possibly,age-related effects on the network level might still be compensated for in LD by some form ofcompensation such as the additional recruitment of domain-general resources [9] subserved bythe networks investigated here, but not in HD conditions. Prediction accuracy for younger adultswas best with fALFF [M = .751] as compared to the other modalities [MGMV = .804, MRSFC =.802, MReHo = .818], while for older adults, the highest accuracy was achieved with GMV [M =.754] as compared to the other modalities [MfALFF = .783, MRSFC = .795, MReHo = .803]. GMV isstrongly linked to atrophy and its pattern of decline is thought to be rather consistent across olderadults [10]. Together with an age-related decrease in performance, this suggests that GMV is amore sensitive marker for individual EF abilities in older adults. For younger adults, fALFF, ameasure of functional variability, might be the more sensitive marker, possibly mediated via itslink to cognitive flexibility [11].Summing up, our results show modality- as well as demand-level specificity. However,overall low brain–behavior associations as well as the missing network specificity, question thepotential of the applied single metrics as markers for individual differences in EF performance.Rather, our findings suggest that future research may need to analyze more global properties ofthe brain, possibly combining different structural and functional metrics or task states as well asage-adapted behavioral testing to result in sensitive predictors for young and old adults,respectively. |