Skip to Content
You are currently on the new version of our website. Access the old version .
  • Proceeding Paper
  • Open Access

18 September 2026

Classification of Stress Levels Using Convolutional Neural Networks and a Wrist-Worn Wearable Device †

,
and
Facultad de Ingeniería y Ciencias Aplicadas, Universidad Central del Ecuador, Quito 170129, Ecuador
*
Author to whom correspondence should be addressed.
†
Presented at the Sixth International Conference on Information Systems and Software Technologies (ICI2ST 2026), Quito, Ecuador, 23–25 September 2026.
Eng. Proc.2026, 153(1), 42;https://doi.org/10.3390/engproc2026153042 
(registering DOI)

Abstract

Stress is a common factor in people’s daily lives. This condition affects health and, consequently, personal, emotional, and work environment. Although stress affects individuals in different ways and at varying levels of intensity, it can lead to a wide range of diseases involving different parts of the human body. Stress is caused by pressures related to past events and concerns about the future, producing symptoms such as anxiety and emotional instability, panic attacks, headaches, insomnia, cardiovascular problems, among others. Monitoring physiological parameters makes it possible to identify risk factors early and thus prevent diseases. In this context, a wrist-worn wearable device captures physiological information through electronic sensors that measure body temperature, heart rate and electrodermal activity. These parameters are correlated to classify stress levels. In recent years, convolutional neural networks have demonstrated the ability to predict and classify several medical conditions using trained models and datasets. Therefore, the aim of this paper is to develop a model based on convolutional neural networks to support the detection of stress levels using wrist-worn wearable device. For this, the research was conducted in two phases: data mining and web development. Regarding data mining, the CRISP-DM methodology was used to define a model through five activities: business understanding, data understanding, data preparation, modeling, and evaluation. The findings indicate that performance metrics derived from the confusion matrix exceeded 90%, including an overall accuracy of 95.47%. Then, the model is implemented within an app using the waterfall model. Conclusions and future work are detailed at the end of the paper.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.