ISPAAD: Integrated Stress, Physical Activity, and Amusement Dataset

Long-term stress exposure is a leading cause of sickness in adults and may even cause physiological issues such as heart diseases. Developing systems to detect and monitor stressful events based on physiological signals can help prevent stress-induced severe health implications. Research in this domain has shown that it is possible to reliably distinguish stress and its absence and differentiate between different types of stress. However, publicly available data in the stress detection domain is rare, which limits the research possibilities. Moreover, datasets are usually very heterogeneous owing to variations in experimental protocols and sensors used to record physiological signals, which limits the comparability. This work presents ISPAAD, an integrated stress, physical activity, and amusement dataset to overcome aforementioned limitations. The dataset is intended for stress detection, in-depth analysis of different stressors, and physical and affective states. The ISPAAD Dataset is provided in 5 different .csv files. The files corresponding to the phyisological signals (e.g., bvp, eda, acc, temp) each contain one sensor measurement per row as well as an identifier and the source dataset. The file corresponding to the labels includes the different interventions (e.g. baseline, cognitive stress, socio-evaluative stress) along with timestamps. The labels file includes the old identifiers from the data sources as well as the ones, created in the course of integrating the different datasets The sources of the data are: -WESAD (https://ubi29.informatik.uni-siegen.de/usi/data_wesad.html) -Wearable Dataset (https://doi.org/10.13026/zzf8-xv61) -VitaStress (https://github.com/paulvincenz/VitaStress)