A Python and scikit-learn research pipeline for classifying frontotemporal dementia patients and healthy controls from resting-state fMRI. The study compared SVM and Random Forest across five functional parameters and three scaling strategies, with grid-search tuning and 100 stratified train/test splits. Reported results were 87% Random Forest accuracy on VMHC and 80% overall, with VMHC four percentage points above the previous published benchmark. The work was published at KES 2022 (Elsevier).
0 likes · 1 view · Nov 5, 2026
A Python and scikit-learn research pipeline for classifying frontotemporal dementia patients and healthy controls from resting-state fMRI.
