Evaluation of Commercial-Off-The-Shelf Wrist Wearables to Estimate Stress on Students
June 16th, 2018
A protocol to evaluate solutions based on commercial-off-the-shelf (COTS) wrist wearables to estimate stress in students is proposed. The protocol is carried out in two phases, an initial laboratory-based stress induction test, and a monitoring stage taking place in the classroom while the student is performing academic activities.
Wearable devices have become very popular. Among them, the most used are wrist wearables such as smart bands or smart watches. These devices are frequently used by sports people as activity trackers for monitoring and tracking fitness-related metrics.
Sensors available in these devices allow to collect data such as users to your location, movement patterns, steps taken, or heart rate, among others. This data can be processed using data analytics techniques to estimate the value of variables such as distance walked, calories consumed, maximal aerobic capacity, or sleep quality. These devices are starting to be introduced in other domains, such as health or education.
In our case, we decided to explore the possibility of using commercial wrist wearers in an academic context. More specifically, our work focuses on estimating the stressing of students using these devices. For this, we measure variations in physiological signs, such as heart rate, skin temperature, or sweating, during the performance of different training activities.
Then, material learning techniques are applied to these signs to estimate stress levels. Commercial off-the-shelf wrist wearable devices have sensors that provide information on physiological signals that have been widely used by the scientific community in stress assessment and detection. Among these signals, heart rate, respiratory rate, skin temperature, or galvanic skin response are the most common.
To carry out this study, an evaluation protocol has been defined, in which the following tools are used. A commercial wrist wearable that is provided to the student to capture physiological signals, a smartphone running an application developed to collect data and submit it to an analysis server, a dashboard that facilitates the visualization of collected data, and finally, the data analytics package that supports the execution of machine learning algorithms. The protocol that will help to stimulate the stress is divided into two phases.
First, a calibration process is carried out in a laboratory, according to fully-controlled conditions. For this, different levels of stress are induced in the student by means of normalized tests, which are the Stroop color and word test, the Paced Auditory Serial Addition Test, and a hyperventilation test. Changes in biological signals are analyzed in these controlled conditions, to be used as a reference for the next phase.
The first task is to create a baseline for a stress analysis. In this task, the student visualizes a four-minute relaxing video, in which different shots of a sunset and bridge and so on. Then the student will be questioned about their perceived stress level.
The second task is an adaption of the Stroop color and word test. In our case, this test is divided into three different levels of difficulty. For this task, the user must choose the color in which the name of a color is painted.
Several buttons located at the bottom of the screen containing the initial letter of a color are available fore the user to choose the color. For example, the button that refers to blue, that picks letter B.Difficulty increases for each level. To increase the student's frustration and increase their stress level, every time an error is committed, an annoying beep will sound.
And if two errors are made, the correct color count will be reset. For the last, most difficult level, name and color will not match. Once each of the three levels is completed, the student will be questioned about their perceived stress level.
The third task consists on the Paced Auditory Serial Addition Test, which measures how the student experiences concentration. During this task, a sequence of consecutive numbers is played aloud, and the student must add the last two numbers and write the result in the provided onscreen box, before listening to the next number. Again, if three errors are committed, the score will be reset.
The student will be questioned about their perceived stress level upon completion of the task. The fourth and the last test consists on a hyperventilation activity to induce the same variation in the physiological signals that would provoke a stressful situation. The student will be inquired about their perceived stress level.
The second phase takes part in the academic room. The objective in this case is to collect the physiological signals of the students while they are performing educational activities. For example, changes in the physiological signals are measured when the teacher asks a question or when the student is completing an answer.
Connect your smartphone and tablet device to a stable internet connection. Turn on Bluetooth communications in the smartphone. On the smartphone and tablet, download and install the official wrist wearable applications for physiological signals and stress test applications.
Turn on the wrist wearable device and place the wearable. In the smart phone, open the official wrist wearable application. The application will proceed to synchronize the wearable device with the smartphone.
For some devices, an email address is required. In the smartphone, open the physiological signals application. Wait for the physiological signals app to display the word Weared"in green.
This indicates that the wearable device has been detected, and therefore the transmission of information from the sensors to the smartphone is enabled. Choose a comfortable and non-disturbing room. Turn on the wrist wearable device.
Place it around the subject's non-dominant wrist, and place the headphones on the head of the subject. Fit tightly but comfortably, the wearable around the wrist. In the smartphone, launch the physiological signals application.
Select the change user option in the left configuration menu, and provide the ID of the subject who will complete the test, and click save. In a laptop, access the dashboard and enter the test administrator's ID and password. Select the subject's ID and select the subject's stress step.
Check the physiological signal's evolution, and wait for the wearable device to reach thermal stability to start the experiment. On the tablet, launch the stress test application. Explain to the subject the four laboratory tests.
Show some of the screens and actions to perform during each of the tasks. This is very important, because the subject shall feel stressed or relaxed in accordance with the performed activities, and not because they experience fear or concern about what is going to occur. Enter the same user ID as in the physiological signals app, and click the arrow.
Launch the video task and give full control to the student. When the task is finished, check that the subject provides the perceived stress. Launch the Stroop color task consecutively for levels one, two, and three.
Only for level three, in case the subject does not solve it after four minutes, terminate the task by pressing the arrow located at the top of the screen. Launch the PASAT test. In case the subject does not solve the test after four minutes, terminate the task by pressing the arrow located at the top of the screen.
Launch the hyperventilation test. Observe the evolution of heart rate, using the dashboard. If the physiological signals do not change significantly, increase inspiration and expiration rates gradually.
In case the subject feels dizziness or uncomfortable, halt the task. In any case, complete the task after four minutes. Turn on the wrist wearable device and place the wearable.
Launch the physiological signals application and choose the correct subject ID.in a laptop, access the dashboard and enter the test administrator's ID and password. Select the subject ID and select the subject's stress stop. Check the evolution of physiological signals.
Take notations about any relevant event occurring in the classroom in relation to student-teacher interaction. Relevant information and basic events will be used to label physiological signal samples afterwards. Example events are a question from the teacher to the student or a theoretical explanation.
At the end of the lecture, ask the subject to complete the questionnaire about their level of stress at the specific times during the session, according to a five level scale. In a laptop, access the dashboard and enter the test administrator's ID and password. Select the subject ID and select the subject's stress step.
Select the day of the classroom experiment. Begin to file lecture room activities and their duration according to the starting and finishing times and their types. For each activity, select a perceived stress level.
For each subject and each session, download the file with the text samples. A comma-separated value file is created for each student. It shows reflecting the values of the physiological signals with their standard deviation, slope, and difference with higher values, the activity type, the activity-based stress, and the subject's perceived stress.
Launch the data analytics package. Choose a set of classifiers. For example, support vector that changed, and import the comma-separated value file for all students for each session.
Train and evaluate a set of classifiers, using the 10-fold cross-validation technique. Finally, check the results of accuracy and error rates. The protocol discussed so far has been validated in a real-world scenario, using a course in computer architectures.
Preliminary results show strong variations in physiological signal values captured in the laboratory, from relaxed video watching to the hyperventilation test. These variations are a consequence of the evolution of the students in the study from the relaxed status to a stressed one. In this context, machine learning classification provides success rates about 90%Success rate worsens when data is captured in the lecture room.
Nevertheless, relevant differences are found when the student is subjected to particularly stressful situations, such as pop quizzes or finals. It is well known that stress may have a most relevant influence in the development of academic activities and overall students'performance. Higher stress levels are especially relevant during the freshman year, with estimated dropout rates between 20%and 30%The proposed protocol is intended to serve as an instrument to eventually define richer student models than those presently used in learning management systems or student information systems.
We think that the student model, enhanced with information of their mental or physical states, will pave the way for new added-value applications and services to improve academic performance. For example, the new information captured with the most suitable wearable device according the protocol discussed, could be applied to the early detection of situations affecting performance, such as fatigue, lack of attention, or stress. And to guide students to overcome those situations.
View the full transcript and gain access to thousands of scientific videos
Overview
This article presents a protocol for evaluating commercial-off-the-shelf wrist wearables to estimate stress levels in students. The study is conducted in two phases: an initial laboratory-based stress induction test followed by a monitoring stage in the classroom during academic activities.
Key Study Components
Area of Science
- Wearable technology
- Stress measurement
- Educational psychology
Background
- Wrist wearables are increasingly popular for tracking fitness metrics.
- These devices collect various data, including heart rate and movement patterns.
- There is growing interest in applying these technologies in educational settings.
- Understanding stress in students is crucial for academic performance.
Purpose of Study
- To evaluate the effectiveness of wrist wearables in estimating student stress.
- To develop a protocol that can be utilized in educational environments.
- To enhance the understanding of stress management in students.
Methods Used
- Conducting a laboratory-based stress induction test.
- Monitoring students in a classroom setting.
- Using data analytics techniques to process collected data.
- Comparing stress levels before and during academic activities.
Main Results
- The protocol effectively identifies stress levels in students.
- Wrist wearables provide reliable data for stress estimation.
- Classroom monitoring reveals insights into stress during academic tasks.
- Findings support the integration of wearables in educational settings.
Conclusions
- Wrist wearables can be valuable tools for monitoring student stress.
- The proposed protocol can be adapted for broader educational research.
- Further studies are needed to explore long-term effects of stress on learning.
Frequently Asked Questions
Executive Industry Relevance
Estimating student stress using commercial-off-the-shelf wrist wearables offers a scalable approach to monitor physiological responses in educational settings, supporting early detection of cognitive load and attentional states. This protocol enables biopharma R&D teams to validate wearable-derived biomarkers for stress-related phenotypes in preclinical and clinical models, particularly for neuropsychiatric indications where stress modulation is a therapeutic target. By integrating quantitative sensor data with qualitative self-reports, the method enhances predictive confidence in translational models linking physiological markers to behavioral outcomes.
Strategic Applications in Biopharma R&D
Early Discovery & Target Validation
- Scientific Value: Enables interrogation of stress-response pathways through longitudinal physiological monitoring in disease-relevant systems.
- Operational Value: Supports target validation by correlating wearable-derived heart rate variability and galvanic skin response with pharmacological interventions in stress models.
- Predictive Value: Improves confidence in lead compounds by providing objective, quantifiable stress metrics that reduce reliance on subjective endpoints.
Screening & Assay Development
- Scientific Value: Prepares validated physiological readouts for high-throughput screening of compounds targeting stress modulation.
- Operational Value: Standardizes assay conditions using controlled stress induction protocols (Stroop, PASAT, hyperventilation) to ensure reproducibility across screening campaigns.
- Scalability: Leverages widely available COTS devices, enabling decentralized data collection and reducing dependency on specialized clinical equipment.
Translational & Preclinical Research
- Scientific Value: Bridges discovery and preclinical validation by aligning wearable biomarkers with clinical stress assessments in humanized models.
- Operational Value: Facilitates continuous monitoring in freely moving subjects, improving ecological validity of preclinical stress studies.
- Risk Mitigation: Enables early identification of compounds with off-target stress-inducing effects, supporting de-risking in CNS and metabolic disease programs.
Pipeline & Workflow Integration
The protocol fits within the discovery continuum from target validation through lead optimization, where physiological stress metrics inform mechanism of action and safety profiling.
- Discovery Biology: Supports hypothesis testing of stress-related targets by providing time-resolved physiological data during controlled challenges.
- Screening: Delivers quantitative, multi-parametric outputs (heart rate, skin temperature, galvanic skin response) suitable for assay readiness and hit confirmation.
- Analytics: Applies machine learning classification to multivariate sensor data, enabling pattern recognition and stress-state discrimination.
- Translational Research: Connects laboratory-induced stress responses to real-world academic stressors, modeling clinical relevance of biomarker trajectories.
- Enterprise Reuse: Establishes a reusable framework for stress biomarker validation across therapeutic areas, promoting cross-platform consistency.
Operational & Enterprise Impact
- Scientific Value: Increases predictive confidence in target engagement by reducing noise in stress-related phenotypic readouts.
- Operational Value: Enhances reproducibility through standardized stress induction and synchronized multimodal data capture.
- Strategic Value: Improves go/no-go decisions by identifying compounds that exacerbate or mitigate physiological stress signatures early in development.
- Portfolio Impact: Enables risk-adjusted prioritization of candidates based on their impact on stress homeostasis, reducing late-stage attrition in neuropsychiatric indications.
Implementation Considerations
- Requires expertise in physiological signal processing and machine learning for multivariate time-series analysis.
- Depends on compatible COTS wrist wearables with access to raw sensor streams (PPG, temperature, GSR) and stable Bluetooth connectivity.
- Necessitates cross-team alignment between data scientists, pharmacologists, and behavioral scientists for synchronized data labeling and interpretation.
- Must account for inter-device variability in sensor accuracy and sampling rates when scaling across models or sites.
- Limited by motion artifacts during active behaviors, requiring artifact correction protocols or sedentary challenge designs for reliable stress estimation.
