Aim of the paper is to illustrate how wearable body sensor networks (BSNs) powered by Machine Learning (ML) algorithms and a suitable visualization tool may be used to timely discover risky situations due to high mental stress. Although many algorithms and wearable devices are able to measure the stress, algorithms predictive of a trend towards, or pointing out timely, risky stress situations need the availability of sequences in time of multiple sensed data depending on context. The paper shows how this can be obtained by means of a wearable body sensor network and two cooperating neural networks: one resident on the edge devices and the other on a remote control center. The former to test the current situation and to possibly alert assistance people, the latter to learn the mental stress model and to predict risky stress situations for the subject under test. Recognition of the incipience of a risky situation is obtained by visualizing how the stress status evolves in the Rn space containing the Rn stress feature clustering by means of its projections into the Rn-1 using the Schlegel diagram. The case study illustrates how the mental stress prognostics may be done by visualizing the projections in the 3D Schlegel diagram of a 4D hypercube (called tesseract) containing the stress features clustered in a R4 space.