A Wearable Exam Stress Dataset for Predicting Grades using Physiological Signals

Md Rafiul Amin, Dilranjan S. Wickramasuriya, Rose T. Faghih

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    The study of psychological stress in real-world scenarios presents several challenges. Consequently, datasets available to researchers are also scarce. The aim of our study is to acquire such a dataset containing skin conductance measurements and use it to predict human performance. We collected skin conductance and skin temperature data from 10 subjects during three exams using wearable devices. We filter the skin conductance signals to obtain coarse-grained trendlines and then train classifiers to predict high and low grades based on the trendline features. We obtained maximum classification accuracies in the 70–80% range. We also obtained the mean trendlines indicating the general variation of stress levels during the exams. The findings indicate the preliminary viability of using wearable devices to predict performance during real-world stressors. Wearable monitoring presents unique challenges and it is our hope that this publicly-available dataset will aid in addressing some of them.
    Original languageEnglish (US)
    Title of host publicationHealthcare Innovations and Point of Care Technologies Conference, HI-POCT 2022
    Pages30-36
    Number of pages7
    ISBN (Electronic)9781665496155
    DOIs
    StatePublished - 2022

    Publication series

    NameHealthcare Innovations and Point of Care Technologies Conference, HI-POCT 2022

    Fingerprint

    Dive into the research topics of 'A Wearable Exam Stress Dataset for Predicting Grades using Physiological Signals'. Together they form a unique fingerprint.

    Cite this