Novel Approach for Stress Detection Using Smartphone and E4 Device
30 Jul 2019-pp 736-745
TL;DR: A novel methodology is proposed by creating a personalized model from the generalized model for stress detection to give more accuracy as two devices are used with the novel approach of model building.
Abstract: Stress reduction is important for maintaining overall human health. There are different methodologies for detecting stress including clinical tests, traditional methods and various sensors and systems developed using either a smartphone, wearable devices or sensors connected to the human body. In this paper, a novel methodology is proposed by creating a personalized model from the generalized model because stress differs from person to person for the same work profile. A generalized model for stress detection is developed from smartphone and E4 device data of all the available individuals. A generalized model is used to build a personalized model that is a person-specific model and will be build up over a time of time when enough amount of person-specific data gets collected. This proposed methodology intends to give more accuracy as two devices are used with the novel approach of model building. Various machine learning algorithms such as ANN, xgboost, and SVM are implemented with the E4 device dataset while the LASSO regression model is used for smartphone data. ANN worked best than xgboost and SVM with 93.71% accuracy. In LASSO, 0.6556 RMSE is achieved.
Citations
More filters
TL;DR: Wang et al. as discussed by the authors designed two classes of controllers: (1) an inhibitory controller for reducing cognitive stress and (2) an excitatory controller for increasing cognitive stress.
Abstract: Keeping cognitive stress at a healthy range can improve the overall quality of life: helping subjects to decrease their high levels of arousal, which will make them relaxed, and elevate their low levels of arousal, which could increase their engagement. With recent advances in wearable technologies, collected skin conductance data provides us with valuable information regarding ones’ cognitive stress-related state. In this research, we aim to create a simulation environment to control a cognitive stress-related state in a closed-loop manner. Toward this goal, by analyzing the collected skin conductance data from different subjects, we model skin conductance response events as a function of simulated environmental stimuli associated with cognitive stress and relaxation. Then, we estimate the hidden stress-related state by employing Bayesian filtering. Finally, we design a fuzzy control structure to close the loop in the simulation environment. Particularly, we design two classes of controllers: (1) an inhibitory controller for reducing cognitive stress and (2) an excitatory controller for increasing cognitive stress. We extend our previous work by implementing the proposed approach on multiple subjects’ profiles. Final results confirm that our simulated skin conductance responses are in agreement with experimental data. In a simulation study based on experimental data, we illustrate the feasibility of designing both excitatory and inhibitory closed-loop wearable-machine interface architectures to regulate the estimated cognitive stress state. Due to the increased ubiquity of wearable devices capable of measuring cognitive stress-related variables, the proposed architecture is an initial step to treating cognitive disorders using non-invasive brain state decoding.
8 citations
27 Jun 2020
TL;DR: This paper mainly investigates whether physiological data can be considered and used as a form of implicit user feedback, and highlights the importance of having a context analyzer, which can help the system to determine whether the detected stress could be considered as actionable and consequently as implicituser feedback.
Abstract: Ensuring the quality of user experience is very important for increasing the acceptance likelihood of software applications, which can be affected by several contextual factors that continuously change over time (e.g., emotional state of end-user). Due to these changes in the context, software continually needs to adapt for delivering software services that can satisfy user needs. However, to achieve this adaptation, it is important to gather and understand the user feedback. In this paper, we mainly investigate whether physiological data can be considered and used as a form of implicit user feedback. To this end, we conducted a case study involving a tourist traveling abroad, who used a wearable device for monitoring his physiological data, and a smartphone with a mobile app for reminding him to take his medication on time during four days. Through the case study, we were able to identify some factors and activities as emotional triggers, which were used for understanding the user context. Our results highlight the importance of having a context analyzer, which can help the system to determine whether the detected stress could be considered as actionable and consequently as implicit user feedback.
3 citations
References
More filters
08 Jan 2018
TL;DR: A novel research problem that tries to detect the compound emotion of smartphone users, observing that users' self-reported emotional states have high correlation with their smartphone usage patterns and sensing data and proposing a machine learning algorithm for compound emotion detection based on the smartphone sensing data.
Abstract: Social psychology and neuroscience had confirmed that emotion state exerts a significant effect on human communication, perception, social behavior and decision making. With the wide availability of smartphones equipped with microphone, accelerometer, GPS, and other source of sensors, it is worthwhile to explore the possibility of automatic emotion detection via smartphone sensing. Particularly, we focus on a novel research problem that tries to detect the compound emotion (a set of multiple dimensional basic emotions) of smartphone users. We observe that users' self-reported emotional states have high correlation with their smartphone usage patterns and sensing data. Based on the observations, we exploit a feature extraction and selection algorithm to find the most significant features. We further adopt a factor graph model to tackle the correlations between features and emotion labels, and propose a machine learning algorithm for compound emotion detection based on the smartphone sensing data. The proposed mechanism is implemented as an APP called MoodExplorer in Android platform. Extensive experiments conducted on the smartphone data collected from 30 university students show that MoodExplorer can recognize users' compound emotions with 76.0% exact match on average.
67 citations
15 Jul 2015
TL;DR: The findings show that the perceived stress is highly subjective and that only person-specific models are substantially better than the baseline.The goal is to develop a machine-learning model that can unobtrusively detect the stress level in students using data from several smartphone sources.
Abstract: This paper presents an approach to detecting perceived stress in students using data collected with smartphones. The goal is to develop a machine-learning model that can unobtrusively detect the stress level in students using data from several smartphone sources: accelerometers, audio recorder, GPS, Wi-Fi, call log and light sensor. From these, features were constructed describing the students' deviation from usual behaviour. As ground truth, we used the data obtained from stress level questionnaires with three possible stress levels: "Not stressed", "Slightly stressed" and "Stressed". Several machine learning approaches were tested: a general models for all the students, models for cluster of similar students, and student-specific models. Our findings show that the perceived stress is highly subjective and that only person-specific models are substantially better than the baseline.
66 citations
TL;DR: Physiological signals such as electrodermal activity and heart rate can help computing systems detect a user's stress level, but integrating additional measures not yet exploited could significantly increase stress-detection accuracy, enabling many new applications.
Abstract: Physiological signals such as electrodermal activity and heart rate can help computing systems detect a user's stress level. Integrating additional measures not yet exploited by such systems could significantly increase stress-detection accuracy, enabling many new applications.
58 citations
26 Mar 2018
TL;DR: This work proposes a so-called heterogeneous multi-task feature learning method that jointly builds inference models for related tasks but of different types including classification and regression tasks and identified strong depression indicators such as time staying at home and total time asleep.
Abstract: Depression is a common mood disorder that causes severe medical problems and interferes negatively with daily life. Identifying human behavior patterns that are predictive or indicative of depressive disorder is important. Clinical diagnosis of depression relies on costly clinician assessment using survey instruments which may not objectively reflect the fluctuation of daily behavior. Self-administered surveys, such as the Quick Inventory of Depressive Symptomatology (QIDS) commonly used to monitor depression, may show disparities from clinical decision. Smartphones provide easy access to many behavioral parameters, and Fitbit wrist bands are becoming another important tool to assess variables such as heart rates and sleep efficiency that are complementary to smartphone sensors. However, data used to identify depression indicators have been limited to a single platform either iPhone, or Android, or Fitbit alone due to the variation in their methods of data collection. The present work represents a large-scale effort to collect and integrate data from mobile phones, wearable devices, and self reports in depression analysis by designing a new machine learning approach. This approach constructs sparse mappings from sensing variables collected by various tools to two separate targets: self-reported QIDS scores and clinical assessment of depression severity. We propose a so-called heterogeneous multi-task feature learning method that jointly builds inference models for related tasks but of different types including classification and regression tasks. The proposed method was evaluated using data collected from 103 college students and could predict the QIDS score with an R2 reaching 0.44 and depression severity with an F1-score as high as 0.77. By imposing appropriate regularizers, our approach identified strong depression indicators such as time staying at home and total time asleep.
58 citations
TL;DR: A comprehensive framework for the early detection of mental stress is proposed by analysing variations in both electroencephalogram (EEG) and electrocardiogram (ECG) signals from 22 male subjects and indicates significant neurophysiological differences between the stress and control (stress-free) conditions at the individual level.
55 citations