ethan whitman
ethanwhitman777.bsky.social
ethan whitman
@ethanwhitman777.bsky.social
phd student at duke with moffitt/caspi + laboratory of neurogenetics using predictive modeling to understand brain aging
yes definitely!!
October 4, 2025 at 8:11 PM
Congrats, Armin!!!
October 3, 2025 at 11:25 AM
thanks ted!!!
July 16, 2025 at 5:54 PM
Many thanks to my co-first author @maxwellelliott.bsky.social, Annchen Knodt, our awesome team at Duke and Otago, and my advisors Av Caspi, Terrie Moffitt, and Ahmad Hariri!
July 1, 2025 at 2:59 PM
We want this measure to be available to the scientific community to help uncover how aging is related to disease, exposure, and interventions.

The algorithm to estimate DunedinPACNI is publicly available at github.com/etw11/Dunedi...
GitHub - etw11/DunedinPACNI: Code to estimate DunedinPACNI scores from FreeSurfer parcellations of brain MRI data.
Code to estimate DunedinPACNI scores from FreeSurfer parcellations of brain MRI data. - etw11/DunedinPACNI
github.com
July 1, 2025 at 2:59 PM
DunedinPACNI is a novel, distinct approach to measuring aging from neuroimaging that can be easily estimated from simple measures from T1 MRI data.
July 1, 2025 at 2:59 PM
DunedinPACNI tended to explain a bit more variance in clinical outcomes. Notably, this variance did not overlap with brain age gap very much and DunedinPACNI and brain age gap were not very correlated.
July 1, 2025 at 2:59 PM
Lastly, we wanted to test whether DunedinPACNI was distinct from brain age gap, a neuroimaging biomarker trained on chronological age.
July 1, 2025 at 2:59 PM
DunedinPACNI’s association with cognitive impairment in BrainLat was similar to that in ADNI, an initial indication of good generalization among people from Latin America.
July 1, 2025 at 2:59 PM
Brain-based predictive models often fail when there are demographic differences between training and test samples – posing a hurdle for clinical translation. We tested if this was true for DunedinPACNI using the BrainLat dataset, a sample of Latin American adults with and without dementia.
July 1, 2025 at 2:59 PM
In UK Biobank, faster DunedinPACNI scores were related to frailty, number of chronic illnesses, risk of new chronic illness, and risk of death from any cause. Thus, DunedinPACNI appears to index aging across the entire body
July 1, 2025 at 2:59 PM
So, DunedinPACNI seems like a useful predictor of cognitive and brain decline. But what about aging more broadly?
July 1, 2025 at 2:59 PM
Furthermore, faster DunedinPACNI scores at baseline predicted more rapid hippocampal atrophy in the future.
July 1, 2025 at 2:59 PM
In ADNI and UK Biobank, faster DunedinPACNI scores were correlated with worse cognition, cognitive impairment, and risk of cognitive decline
July 1, 2025 at 2:59 PM
Using Dunedin Study data, we trained an algorithm to estimate the Pace of Aging from brain structure. We call this measure “DunedinPACNI” – Pace of Aging Calculated from NeuroImaging

Then we applied this algorithm to external data to test its association with clinical outcomes.
July 1, 2025 at 2:59 PM