Predicting cognitive decline from non-brain and neuroimaging-based data published on Neurobiology of Aging 2022-07-23
Very pleased to announce that our work "Predicting future cognitive decline from non-brain and multimodal brain imaging data in healthy and pathological aging" has been published in Neurobiology of Aging! This is the first work published during my postdoc. We were added later into the project to continue it, and worked to see it through the peer-review process, addressing reviewer concerns and comments.
Read more in:
Predicting future cognitive decline from non-brain and multimodal brain imaging data in healthy and pathological aging, Neurobiology of Aging, 2022-07-22, doi: 10.1016/j.neurobiolaging.2022.06.008
Or get the TLDR on Twitter:
It's with great pleasure that we present "Predicting future cognitive decline from non-brain and multimodal brain imaging data in healthy and pathological aging", now published in Neurobiology of Aging! 🧵1/12 https://t.co/ZJ9lsI4GO3— Bruno Hebling Vieira (@HeblingVieira) July 23, 2022
In this work, we propose a new method to predict future cognitive decline from multimodal brain imaging data in healthy and pathological aging. Using data from the OASIS-3 dataset, we predict yearly rate of change in MMSE and CDR-SOB scores. A multi-target Random Forest regression model with multivariate imputation was trained and evaluated for this task. We show that non-brain data (demographics, clinical and neuropsychological scores) are strong predictors of cognitive decline. While brain morphometry demonstrated incremental improvements in the predictive power of our model, we also found that functional connectivity did not demonstrate improvements. The most important features of our model are neuropsychological scores (MMSE and CDR-SOB), and subcortical volumes, assessed at the baseline date.
We hope that these results help to develop brain-based markers of cognitive decline, which can be useful in the future for early-detection and the development of interventions and treatments, not only for the pathological population but applied to healthy aging as well.
I would like to thank the coauthors for the invitation to contribute to this project and the opportunity to learn better-practices. I hope you enjoy the read!