Student Posters Repository

25% of all mental diseases consist of types of depression, anxiety disorder and obsessive-compulsive disorder. Due to the limited treatment options with regard to these mental diseases compared to their prevalence, duration of the disease is often prolonged. To fill this gap, we aim at detecting typical movement patterns indicating an episode of the diagnosed mental disease and implementing this onto a wearable device carried during normal day activity, allowing for direct feedback to the patient.

To do so, we develop a classification system for human motion that allows us to detect those patterns directly in a data stream of several sensors. In addition to that, we are interested in reconstructing those patient motions in order to yield insight into the typical patterns exhibited by patients. This, however, depends significantly onto the quality and the amount of the data, therefore we aim at building an optimization system that yields the optimal placement of such sensors in order to maximize reconstruction accuracy.