Human Stress Analysis using Physiological Signals from Wireless Healthcare Sensor Network

Abstract

Stress and emotion are inter-related and contribute to mental health. Research suggests chronic stress can impact on the biological, psychological, and behavioral responses to stress in life. For analysis of stress, physiologists study and analyze the physiology of human stress response. With the advent of information and communication technology
and wireless sensor network (WSN), and wearable sensors, the paradigm of e-health services has been changed, and emphasized the importance of being personalized. A variant of wireless body area network is proposed which is used as wireless healthcare sensor network (WHSN) for collecting physiological signals from human body using different sensors and analyzed for stress levels.
A unimodal stress detection technique using galvanic skin response (GSR) signal in the WHSN framework is analyzed through a prototype developed using WSN and GSR sensor. Using supervised classifiers, the stress levels are classified using the features generated from the GSR signal after preprocessing. A bimodal stress detection technique using GSR and photoplethysmogram (PPG) signals is proposed and prototype experimental setup is done using WHSN node, GSR, and PPG sensors. With the help of supervised classifiers, the stress levels are classified using the features generated from the GSR and PPG signals after preprocessing. Three different multimodal stress detection/analysis approaches using different combination of three or more signals such as GSR, PPG, respiration inductive plethysmogram (RIP), and skin temperature (ST) are proposed. Features are generated from these signals after preprocessing and classified using supervised classifiers for stress level detection. Along with our sample dataset, a publicly available dataset (BNCI Horizon 2020) is used for validation of our stress detection techniques and the results are presented.
The patient data are always critical, and should be stored and accessed in such a way that it should be accessed by only authorized users such as medical practitioners or healthcare providers. So, user authentication is a major concern in an electronic healthcare system. A fingerprint-based anonymous authentication technique between the user and medical application server in WHSN is proposed, which preserves patient’s privacy as well as achieves mutual authentication. The security analysis of the proposed protocol is discussed. The proposed protocol resists well-known attacks such as replay, forgery, privileged insider, and stolen verifier attacks. Comparison in terms of computation overhead and time with three existing schemes shows that the proposed scheme is light-weight and is suitable for WHSN based healthcare systems.