Skip to Main Content
We perform prediction of diverse disorders (cocaine use, schizophrenia and Alzheimer's disease) in unseen subjects from brain functional magnetic resonance imaging. First, we show that for multisubject prediction of simple cognitive states (e.g., motor versus calculation and reading), voxels-as-features methods produce clusters that are similar for different leave-one-subject-out folds; while for group classification (e.g., cocaine addicted versus control subjects), voxels are scattered and less stable. Therefore, we chose to use a single region per experimental condition and a majority vote classifier. Interestingly, our method outperforms state-of-the-art techniques. Our method can integrate multiple experimental conditions and successfully predict disorders in unseen subjects (leave-one-subject-out generalization accuracy: 89.3% and 90.9% for cocaine use, 96.4% for schizophrenia and 81.5% for Alzheimer's disease). Our experimental results not only span diverse disorders, but also different experimental designs (block design and event related tasks), facilities, magnetic fields (1.5T, 3T, 4T) and speed of acquisition (interscan interval from 1600 to 3500 ms). We further argue that our method produces a meaningful low-dimensional representation that retains discriminability.