SCZ SVM

Course Project for PSYCH 115 (Brain Imaging Analysis Methods)

Advised by Professor Kevin Weiner, University of California, Berkeley

I investigated whether anatomical features (cortical thickness) or functional features (task-based functional connectivity) are better at classifying schizophrenia through analyzing an open-source fMRI dataset on OpenNeuro. I trained Support Vector Machine (SVM) classifiers using Scikit-learn based on anatomical features and functional features. The model reached a peak accuracy of 84%, and functional features are of greater discriminability than anatomical features.