Unsupervised clustering of neuropsychological tests reveals three cognitive phenotypes in MS with distinct brain atrophy patterns

Three cognitive phenotypes—from preserved cognition to global impairment—were identified via unsupervised clustering of neuropsychological features in 79 MS patients, each associated with different degrees of brain atrophy on MRI.

• Feature selection using variance and mutual information reduced neuropsychological variables, enabling the Partitioning Around Medoids (PAM) algorithm to cluster patients into three distinct groups.
• The resulting phenotypes integrate cognitive performance with structural MRI brain atrophy measures, offering a potential framework for earlier detection of cognitive decline in MS.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.