Continuous Monitoring and Assessment of Mental Workload from Physiological and Activity Data

Type: Semester thesis
Students: Sebastian Steiner, Thomas Steiner
Adviser:  Burcu Cinaz, Bert Arnrich

In a former study we addressed how we can simulate and discriminate different levels of mental workload in a laboratory setting. The aim of this thesis was to extend the lab experiment with a long term monitoring of daily routines by considering different activities occurring during the day. In this study we picked up two subjects who participated in the lab experiment and monitored them in three different days which are characterized by their mental demands (low-medium-high). During daily monitoring, participants labelled their activities by taking pictures using a mobile phone. The recording time was between 3.5 and 5 hours. In order to simulate high workload day, we defined high mental tasks which had to be performed under time pressure and social threat. During the experiments, ECG and the acceleration data were recorded. In order to classify the different workload levels based on heart rate variability, we calculated time- and frequency domain features. Besides, the acceleration data was analysed in order to understand the source of physiological data in an accurate manner. We could discriminate low, medium and high workload days by applying K-means clustering and kNN algorithm using extracted heart rate features.