Abstract : Stress is one of the factors that affect human health in many aspects. It is considered as one of the culprits in increasing the risk of getting sick that could probably lead to critical physical or mental illnesses. Stress can be experienced everywhere and in different circumstances. Hence, stress should be controlled and managed by monitoring its progress or regress. Physiological information can be used to determine stress levels. One of these is the Galvanic Skin Response (GSR) that utilizes skin conductance which is known to be directly involved in the emotional behavioural regulation in humans. In this study, a method on how to determine stress when a person is engaged in mobile communication is proposed. An Android application was developed that is capable of determining the stress level of a person while doing SMS composition. GSR data were utilized and the performance of the proposed method was found of no significant difference with a commercially available device. Factors like phone size and period of texting was investigated and were found out that these only contribute an extremely low level of stress. The developed App could be used to determine stress levels especially if emotional conversations are considered.
 • We aimed to identify and exploit the frequency patterns that relate to the existence of stressful conditions in a participant’s recording. • The existing repository1 was ported to Tensorflow 2.0 and Keras framework in order to work on latest standards and export compatible TensorFlow Lite (an optimized FlatBuffer format) models for the Android platform. • By relying on an objective proxy for stress (GSR) and a robust statistical framework (PSM), this study strengthens substantially the existing literature on mental health benefits of active travel. • Heart rate and GSR can currently be reliably sensed passively in existing wristworn wearables, this paper focuses on using these sensors from a commercially available smartwatch.
 • The preprocessing step is taking as input the raw ECG and EDA and extracts several statistical features for stress detection that have been proposed for physiological signal analysis. • Due to differences in the incorporated datasets, they indicate that the proposed CNN classifier based only on ECG can achieve better or similar performance with the multi-sensor features classified by standard ML techniques (evaluated by 10CV). • Moreover, the proposed CNN-based spectogram analysis revealed that the temporal variation of the spectrum of frequencies seems to have high discriminative power for stress identification. • The problem of stress identification and categorization from the sensor data stream mining perspective, consider a reductionist approach for arousal identification as a drift detection task, highlight the foremost problems of managing with GSR data, and propose simple approaches the way to them.

We have more than 145000 Documents , PPT and Research Papers

Have a question ?

Mail us :