N°386 – Seizure detection using personalized machine learning methods based on wearable ECG

It is previously discovered that evacuees’ mental state involves two key elements: the awareness of the situation, namely the clarity; and the intensity of the stimulation (Deng et al., 2022). In this paper, we aim to construct a clarity-intensity model for the delay time and the speed preference through a VR experiment, and to compare the model with real-world cases. 15 disaster stimuli are designed to build a clarity-intensity plane, and are applied in the virtual environment (VE). The VE is made walkable with an omni-directional treadmill, and the behaviours are videotaped for later analysis. The analysed behaviours are quadratically regressed on the clarity-intensity plane, and it is discovered 1) that the delay time is effectively reduced with high clarity, and appears an inverted U-shape on the intensity dimension, and 2) that the stride frequency ratio, which represents the speed preference, rises with both clarity and intensity. The results on the delay time are further compared with a dormitory building fire case, and consistency is concluded.