Figure 4. a) Connectograms illustrating the relationship between edge-wise variability and age, thresholded at the pnetwork<.05. The same hot-cold color scale limits were applied to all connectograms to facilitate the comparison of the edge variability-age group associations across connectograms. b) cortical surface maps illustrating the relationship between cortical thickness variability and age groups, thresholded at pcluster<.05.
source: Yu, J.. (under review). Age-related decrease in intersubject similarity of cortical morphology and task and resting-state functional connectivity
Figure 2. The top panel illustrates the distribution of prediction metrics of the age-prediction models across 5000 random train-test split iterations: a) Correlation between predicted and actual age. b) Age-prediction bias. c) Mean absolute error. d) Mean absolute error after correcting for the age-prediction bias. The bottom panel depicts the distribution of ridge regression coefficients in the e) cognitive age-prediction and f) physical age-prediction models across 5000 random train-test split iterations. To facilitate interpretations, some of the coefficients for the predictors were sign-flipped (i.e., multiply by -1), such that for all predictors, a positive coefficient would suggest poorer performance corresponding to increasing age. The means of the distributions are illustrated as horizontal and vertical lines in the top and bottom panels, respectively.
source: Yu, J., Ng, T.K.S., & Mahendran, R. (2023) Cognitive and physical age-gaps in relation to mild cognitive impairment and behavioral phenotypes. GeroScience
Figure 4. Beta coefficients predicting ‘UnfamFace_Recog’ scores which were obtained in the gray matter, resting-state functional connectivity, and structural connectivity modalities, averaged across 1000 permuted train-test iterations in the a) young and b) old training samples. For the purpose of illustrating the region-wise resting-state functional and structural connectivity profiles in the chord diagrams, the edge beta coefficients are averaged across their respective regions. Details regarding the mapping of the nodes to their respective brain regions in the functional and structural connectomes are available at https://atlas.brainnetome.org/bnatlas.html and from table 3 of Rolls et al. (2015), respectively.
source: Yu, J., & Fischer, N. L. (2022). Asymmetric generalizability of multimodal brain-behavior associations across age-groups. Human Brain Mapping.
Figure 6. Chord diagrams showing the SC and FC averaged connectivity values between and within different regions, associated with the three behavioral measures with the highest and lowest RMSEold/RMSEyoung ratios in the combined SC and rsFC model. The list of nodes corresponding to various regions in the SC and FC networks can be referred to in table 3 of Rolls et al. (2015) and at https://atlas.brainnetome.org/bnatlas.html, respectively.
Source: Yu, J., & Fischer, N. L. (2022). Age-specificity and generalization of behavior-associated structural and functional networks and their relevance to behavioral domains and imaging modalities. Human Brain Mapping.
Figure 2. a) relationship between depression symptoms and functional connectivity edges. b) Heatmap showing the overall functional connectivity changes (indexed by the number of positive edges – number of negative edges) between and within regions. c) relationship between depression symptoms and structural connectivity edges.
Source: Yu, J., Rawtaer, I., Feng, L., Kua, E. H., & Mahendran, R. (2021). The functional and structural connectomes associated with geriatric depression and anxiety symptoms in mild cognitive impairment: cross-syndrome overlap and generalization. Progress in Neuro-Psychopharmacology & Biological Psychiatry. 110: 110329
Figure 2. Partial correlation networks depicting items from the geriatric depression scale, geriatric anxiety inventory and the friendship scale at (a) Pre-COVID-19 and (b) lockdown. Differences across time in edges values are presented in the (c) Lockdown—Pre-COVID-19 network. Only the edges with significant changes (uncorrected p > 0.05) across time, as determined by the paired network comparison test, are shown. “Positive association” and “Negative association” in the context of (c) meant that the edges became more positive or negative, respectively, across time. Thicker lines corresponded to stronger associations.
Source: Yu, J., & Mahendran, R. (2021). COVID-19 lockdown has altered the dynamics between affective symptoms and social isolation among older adults: results from a longitudinal network analysis. Scientific Reports. 11:14739
Figure 1. Chord diagram illustrating the changes in neurocognitive diagnoses across time among individuals diagnosed with MCI at baseline. The chord diagram is generated using the data reported in a follow-up study (Michaud, Su, Siahpush, & Murman, 2017) of 1,415 MCI subjects (average follow-up = 4.3 years). The width of the links corresponds to the proportion of subjects, within their respective MCI subtypes. The proportion of the sectors representing the MCI subtypes not linked to any subsequent changes in diagnoses correspond to the proportion of subjects remaining at the same diagnosis at follow-up. ‘Other dementias’ include frontotemporal lobe dementia, vascular dementia, Lewy bodies dementia, Parkinson’s disease, and mixed dementias.
Source: Yu, J., Lam, C. L. M., & Lee, T. M. C. (2021). Mild Cognitive Impairment. In G.J Boyle (Ed.), The SAGE Handbook of Clinical Neuropsychology. Sage Publishing.