Involved disciplines Computer Science in PAnalytics
Derivation of new forms of activity recognition at home and in the living environment, as well as development of methods for data analysis
One of the scientific challenges of the computer science lies in the detection of daily routines. Generally, activity recognition is understood as the intention to derive actions and goals of one or more persons from a series of observations like activities and environmental conditions. For this purpose sensor data is processed by means of techniques, for example, from the area of Data Mining or machine learning. By using the right tools it is possible to model a wide range of human activities.
For the use of sensors there are basically three different approaches: wearable sensors (e.g. Samsung Gear), ambient sensors and a combination of both (hybrid approach).
By the use of wearable sensors the interaction environment of the users can be analysed or they can be located, for example. Furthermore, they are given the possibility to communicate with the system on a steady basis. With ambient sensors courses of action of the daily life can be captured, recorded and evaluated.
Health-relevant causal links have to be described in a machine-readable form to identify relationships between daily activities like physical activity, sleep and social interaction. Moreover, metrics have to be developed that formalise health-relevant activity clusters like sleep, social interaction or diet to make them evaluable. These metrics will be used with the causal links mentioned above to generate recommendations for health-enhancing activities.