 Acupuncture has been shown to have beneficial effects on the cognitive function of the brain. To better understand how these effects occur, it is important to be able to differentiate between various manipulations used during acupuncture treatments. We developed a framework that uses supervised isomap and RNN to detect acupuncture manipulations from EEG data. Our approach first extracts a low-dimensional embedding of the brain's dynamic functional network using reconstructed geodesic distances. This embedding reveals strong acupuncture-specific reconfigurations of the brain's topology. Additionally, we find that the distance traveled along this manifold correlates strongly with changes in acupuncture manipulations. Furthermore, we use Takagi Suginokon, TSK, classifiers to identify acupuncture manipulations based on their nonlinear characteristics. Compared to other classifiers, TSK achieves higher accuracy in identifying acupuncture manipulations at 96.71%. These results suggest that our method could provide neural biomarkers for acupuncture physicians to better understand the mechanisms underlying acupuncture's effects on the brain. This article was authored by Kiley, Jung Wang, Shan Shanli, and others. We are article.tv, links in the description below.