Improving brain age estimates with deep learning leads to identification of novel genetic factors associated with brain aging

Abstract

Brain aging trajectories among those of the same chronological age can vary significantly. Statistical models have been created for estimating the apparent age of the brain, or predicted brain age, with imaging data. Recently, convolutional neural networks (CNNs) have shown the potential to more accurately predict brain age. We trained a CNN on 16,998 UK Biobank subjects, and in validation tests found that it was more accurate than a regression model for predicting brain age. A genome-wide association study was conducted on CNN-derived predicted brain age whereby we identified single nucleotide polymorphisms from four independent loci significantly associated with brain aging. One locus has been previously reported to be associated with brain aging. The three other loci were novel. Our results suggest that a more accurate brain age prediction enables the discovery of novel genetic associations, which may be valuable for identifying other lifestyle factors associated with brain aging.Competing Interest StatementThe authors have declared no competing interest.