BrainAgeNeXt

This study introduces BrainAgeNeXt, a convolutional neural network trained on 11,574 MRI scans from 33 datasets to predict brain age across individuals aged 5 to 95, outperforming existing methods with a mean absolute error of 2.78 ± 3.64 years. The model maintained strong performance across varying image qualities, including motion-corrupted and 7T scans. In three longitudinal multiple sclerosis cohorts, brain age was significantly elevated and increased faster in individuals with worsening disability, supporting its use as a prognostic biomarker and potential clinical trial endpoint.