AI and Neurological Rare & Neuroimmune Diseases ── Multiple sclerosis ── 2026-10-26
Multiple sclerosis: treatment, trials and AI
- Abstract
Brain’s structural and functional connectomes (SC and FC) offer insights into the relationship between brain circuits and cognitive functioning in people with multiple sclerosis (MS). However, diffusion and functional MRI, which are used to extract SC and FC, is expensive and time consuming, therefore are not currently used in the clinical setting. Our prior work demonstrated that estimated SC and FC (eSC and eFC) generated using MS lesion masks and artificial intelligence (AI) models can predict disability as effectively as SC and FC derived from diffusion and functional MRI in MS. In this study, we identify the eSC and eFC brain circuits that are associated with baseline and 4-year follow-up cognition and assess the ability of eSC and eFC in predicting cognition in MS. In a cohort of 171 with MS, the Network Modification tool was performed to estimate SC from the MS lesion mask, which is extracted from structural MRI scans. The eSC was then used as an input to Krakencoder, an encoder-decoder model, to estimate FC. The highest accuracy was obtained when predicting follow-up Symbol Digit Modalities Test (SDMT) using regional eSC or eFC, with a median Spearman’s correlation coefficient of 0.58 for eSC and 0.56 for eFC. Decreased eSC and eFC in the cerebellum and increased eFC in the default mode network were associated with lower follow-up SDMT scores. Our findings demonstrate that eSC and eFC derived from clinically acquired MRI and AI models can effectively predict cognition. The use of lesion-based estimates of connectomes may potentially inform clinical trials by identifying the specific brain circuits that could be targeted for the non-invasive treatments aimed at improving cognitive function in MS. Researchers used AI and clinically acquired MRI scans to estimate brain connectivity networks in people with multiple sclerosis, finding these estimates can moderately predict cognitive performance four years later and highlighting key brain circuits involved in cognition.
Journal IF-equivalent: 18.6 (OpenAlex 2-year mean citedness, value as of 2026-10-25, retrieved 2026-10-26; not the official Clarivate IF)Reference: Tozlu C, Ong D, Piccirillo C, Schwartz H, Jaywant A, Nguyen T, et al. Predicting follow-up cognition using estimated structural and functional connectivity networks and artificial intelligence in multiple sclerosis. Nat Commun. 2026 Oct 23 [Epub ahead of print]. doi:10.1038/s41467-026-75827-7.Checked: Abstract only