The team started out by having rookie and intermediate engineering students both take standard tests as well as answer questions about pictures while sitting in an fMRI scanner. From there, they had the algorithm generate "neural scores" that could predict a student's performance. The more certain parts of the brain lit up, the easier it was to tell whether or not a student grasped the concepts at play.
You're not about to get brain scans in between classes, and there are limitations to the existing research. For one, Dartmouth focused on STEM learning -- it's not clear if your brain would react the same way in a literature class. The neural scores also apply only to narrow demonstrations of knowledge. This could, however, help teachers refine their classes by identifying techniques that resonate with most students before exam results come in. Don't be surprised if school is eventually much more engaging.