From multimodal MRI dataset to machine learning classification framework for Long COVID analysis in the brain (v1)

The shared dataset includes a rich collection of brain MRI quantitative maps on a cohort of healthy controls who had not contracted COVID at the time of the scan, people who had COVID prior to the scan and fully recovered, and patients still experiencing Long COVID. The MRI metrics feature biomarkers for iron accumulation, myelin content, inflammation, microstructure, atrophy, and cerebral blood flow. In addition to the data, a framework for machine learning analysis, given any CSV or Excel table of regional values, is provided. A random forest algorithm implementation is provided as a fully functional example, which returns classification scores and plots for a straightforward interpretation of the results.