The potential of learning analytics (LA) in education has been harnessed as a tool for the collection, measurement, and analysis of student data with the intent of understanding and constructing models that enhance learning experiences (Siemens and Long, 2011; Ferguson, 2012; Siemens and Gasevic, 2012). The data from LA have been utilized by various authors to explore and investigate factors that shape learning behavior in technology-mediated spaces (Lin and Hwang, 2018; Ullmann et al., 2019). Procedures carried out by institutions using learning analytics, which are based on clickstream data or the so-called time-stamped digital traces left by students during their interactions with the learning management system (LMS), have been recognized (Sedraz Silva et al., 2018). Given the ability to discern varied and intricate patterns regarding student behavior, the significance of other educational applications is also acknowledged (Cukurova et al., 2020). The potential of LMSs for early detection of students facing challenges in courses, predicting success, and promoting improved outcomes has been highlighted (Tempelaar et al., 2020).

Though LA has advanced the comprehension of students’ learning processes (Avella et al., 2016), its limitations stem from the one-dimensional nature of its data collection (Rodríguez-Triana et al., 2017). Crucial contextual student data, vital for understanding behaviors and influential factors in their learning process, are often omitted (Eradze and Laanpere, 2017). To address these limitations, multimodal learning analytics (MMLA) has been introduced as a comprehensive approach to data collection (Giannakos et al., 2022; Monsalves et al., 2023). MMLA is characterized by the study of data from diverse sources stemming from student interactions in a range of environments beyond just the digital; this includes physical, physiological, psychometric, and environmental spaces (Worsley et al., 2016; Mu et al., 2020; Giannakos et al., 2022). Such data are captured, amalgamated, and analyzed to provide a holistic understanding of factors associated with the learning process (Quadri and Shukor, 2021).

Specifically, technological advancements have paved the way for wearable devices in the realm of capturing multimodal data pertaining to the physiological environment. These advancements have also simplified the storage of vast amounts of data in the cloud, easing their collection and subsequent analysis (DiMitri et al., 2018). Wearable devices, especially multimodal computerized biofeedback wristbands, have become pivotal for real-time data collection (Garbarino et al., 2014; Regalia et al., 2019). Their widespread adoption is attributed to their increasing popularity in health tracking and monitoring (Vos et al., 2023). Notably, the Empatica E4 wristband, known for its noninvasive nature, cost-effectiveness, and cloud storage capabilities, is prominently utilized (Ollander et al., 2016; Koskimäki et al., 2017). This device has been employed in diverse studies encompassing music, stress, and emotion management (Borrego et al., 2019; Bulagang et al., 2021; Chandra et al., 2021; Rahman et al., 2021).

Recently, there has been considerable interest in creating multimodal systems that automatically analyze students’ states that are hard to observe directly, like cognitive load and stress levels, but can significantly affect their performance (Bustos-López et al., 2022). Some studies have focused on examining aspects such as self-regulation (Spann et al., 2017), motivation (Järvenoja et al., 2018), interest, and attention during activities (Tan et al., 2021). At this point, biometric data aids in managing learning-related information, such as assessment load (Hernandez-de-Menendez et al., 2021). Classroom biometric monitoring can usher in new learning approaches, as timely feedback to students with suitable teaching methods bolsters learning and retention (Ramírez-Moreno et al., 2021a,b).

Consequently, wearable devices provide a range of measurements to acquire biometric data, including physiological responses such as electrodermal activity (EDA), also referred to as galvanic skin response (GSR). This measurement is intimately linked to the sympathetic nervous system (SNS), as skin conductance primarily arises from sweat glands under the exclusive control of the SNS. EDA is widely recognized as a direct method to gauge stress associated with responses of the autonomic nervous system (ANS; Boucsein, 2012). In addition to this, these devices include other peripheral physiological measurements such as respiratory amplitude, skin temperature (ST), electrocardiogram (ECG), blood pressure volume (BVP), muscle electromyography (EMG), and electrooculogram (EOG; Koelstra et al., 2010).

Given the growing interest in wearable devices, it becomes imperative to undertake research in the field of education that addresses both the theoretical and practical issues and the current limitations of these tools. Consequently, this study attempts to pinpoint the characteristics of teaching-learning processes that integrate wearable devices for the collection of biometric data from students. The first section of the article provides an overview of the methodological strategy underpinning this research, based on a systematic literature review (SLR). The following section presents the results of the study, including analysis, interpretation, and findings. The final section addresses the central research question: What are the characteristics of studies that incorporate wearable devices into the analysis of teaching-learning processes?

The systematic review was conducted using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Page et al., 2021), facilitating the identification of potentially relevant articles. Extensive searches were performed in the Scopus and Web of Science (WoS) bibliographic databases to incorporate a wide range of scientific literature while excluding gray literature. Inclusion criteria were established to include peer-reviewed journal articles published in English between 2019 and 2022, spanning quantitative, qualitative, and mixed-methods studies. This approach ensured a comprehensive view of current research integrating wearable devices. Following the delineation of the study objectives, sub-research questions were subsequently defined to answer the main question (Table 1).

The subsequent key findings were derived from the sub-research questions (RQ1). What is the research context that incorporates the use of wearable devices? (RQ2) Are wearable devices utilized to assess the development of any competency component, whether disciplinary or general? (RQ3) Which biometric indicators are associated with the use of wearable devices? (RQ4) What is the methodology for analyzing data gathered via wearable devices?

Wearable devices have been an alternative to identifying performance characteristics in activities contributing to teaching-learning. The study results allowed recognition of the main characteristics of research, such as describing the participants in the activities developed for data collection, the wide range of physiological indicators, and the methodologies to analyze multimodal data. These findings are consistent with those of the study by Giannakos et al. (2022) since they highlight the use of multiple sources of data obtained from the interactions of students using different resources such as videos, e-books, games, and attending face-to-face classes, among others. The wide range of activities associated with collecting multimodal data highlights the relevance of linking and generating multidisciplinary spaces for a better understanding of developing the teaching-learning process in both pedagogical and real-world contexts.

Similarly, a variety of biometric indicators are recognized, including the HR, its variation (HRV), and electrodermal activity (EDA), also known as galvanic skin response (GSR), skin temperature (ST), blood pressure volume (BVP), and accelerometer (ACC), with some being related to the development of disciplinary competencies, such as data estimation, while others focus on generic variables, such as attention, stress, and anxiety. This understanding is in line with the findings described by Hernandez-de-Menendez et al. (2021), who highlighted the potential of biometric data in managing information related to learning. There has been a demonstrated interest in creating multimodal systems that automatically analyze student states that are difficult to observe with the naked eye, such as cognitive load and stress levels (Bustos-López et al., 2022). At this point, the significance of comprehending how data collected through portable devices in pedagogical contexts can serve as a support tool for learning management is emphasized, considering that its collection, fusion, and analysis pose challenges for integration.

This review was initiated with the question, “What are the characteristics of studies that incorporate portable devices in the analysis of teaching-learning processes?” This question led to the development of a methodology that identified the most relevant characteristics responsive to the objectives framed in each investigation. At a general level, the types of participants have been recognized, with university students, graduate students, volunteers, physicians, office and industrial workers, children under 12 years old, children with special educational needs, and patients being among the most prominent ones. It is also acknowledged that these participants were involved in simulated activities, controlled studies, and pedagogical activities aimed at developing data analysis tools and standardized tests. The technological linkage between activities in virtual reality environments and authentic contexts, such as emergencies and courts, is highlighted.

Additionally, other activities encompassed activities of daily living, games that integrate motion-based touchless games (MBTG) with different avatar representations (ASRs), and educational games based on movement (MBEG). Moreover, the trend of utilizing biometric data such as the HR, its variation (HRV), and electrodermal activity (EDA) is highlighted. This is done to measure and detect stress, enable intervention, and improve attention and working memory. Additionally, it emphasizes the easy and portable use of sensors for measuring emotions and physical characteristics. Thus, the usefulness of technology (in this case, wearables) as tools for assessing the impact of an educational intervention is confirmed by our review.

Regarding the limitations of this research, an area of opportunity is identified in research sub-question number four. While the methods of multimodal data analysis are acknowledged, the characteristics considered for applying machine learning techniques to data gathered via wearable devices could be more comprehensively elucidated. Likewise, it is recognized that in this research, it is impossible to demonstrate the direct impact of interventionist improvements in the various integrated contexts. It is proposed that future work should focus on the generation of multidisciplinary teams that can understand performance variables, aiming to create new methods for multimodal data collection and analysis that promote personalized learning. Furthermore, research should be encouraged to identify factors that affect learning to promote more fluid and reflective educational experiences for all those involved in the teaching-learning process.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

LG-M and MC-A contributed to the conception and design of the study. JR-R organized the database. CA-N wrote the first draft of the manuscript. All authors contributed to the article and approved the submitted version.

The authors acknowledge the financial and technical support of Writing Lab, the Institute for the Future of Education, Tecnologico de Monterrey, Mexico, in producing this study. The authors acknowledge the financial support of Tecnologico de Monterrey through the Challenge-Based Research Funding Program 2022. Project ID # I005-IFE001-C2-T3-T.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.