Multimodal Learning Analytics (MMLA) innovations are commonly aimed at supporting learners in physical learning spaces through state-of-the-art sensing technologies and analysis techniques. Although a growing body of MMLA research has demonstrated the potential benefits of sensor-based technologies in education, whether their use can be scalable, sustainable, and ethical remains questionable. Such uncertainty can limit future research and the potential adoption of MMLA by educational stakeholders in authentic learning situations. To address this, we systematically reviewed the methodological, operational, and ethical challenges faced by current MMLA works that can affect the scalability and sustainability of future MMLA innovations. A total of 96 peer-reviewed articles published after 2010 were included. The findings were summarised into three recommendations, including i) improving reporting standards by including sufficient details about sensors, analysis techniques, and the full disclosure of evaluation metrics, ii) fostering interdisciplinary collaborations among experts in learning analytics, software, and hardware engineering to develop affordable sensors and upgrade MMLA innovations that used discontinued technologies, and iii) developing ethical guidelines to address the potential risks of bias, privacy, and equality concerns with using MMLA innovations. Through these future research directions, MMLA can remain relevant and eventually have actual impacts on educational practices.

Simulation-based learning provides students with unique opportunities to develop key procedural and teamwork skills in close-to-authentic physical learning and training environments. Yet, assessing students' performance in such situations can be challenging and mentally exhausting for teachers. Multimodal learning analytics can support the assessment of simulation-based learning by making salient aspects of students' activities visible for evaluation. Although descriptive analytics have been used to study students' motor behaviours in simulation-based learning, their validity and utility for assessing performance remain unclear. This study aims at addressing this knowledge gap by investigating how indoor positioning analytics can be used to generate meaningful insights about students' tasks and collaboration performance in simulation-based learning. We collected and analysed the positioning data of 304 healthcare students, organised in 76 teams, through correlation, predictive and epistemic network analyses. The primary findings were (1) large correlations between students' spatial-procedural behaviours and their group performances; (2) predictive learning analytics that achieved an acceptable level (0.74 AUC) in distinguishing between low-performing and high-performing teams regarding collaboration performance; and (3) epistemic networks that can be used for assessing the behavioural differences across multiple teams. We also present the teachers' qualitative evaluation of the utility of these analytics

No abstract

No abstract

Teacher's in-class positioning and interaction patterns (termed spatial pedagogy) are an essential part of their classroom management and orchestration strategies that can substantially impact students' learning. Yet, effective management of teachers' spatial pedagogy can become increasingly challenging as novel architectural designs, such as open learning spaces, aim to disrupt teaching conventions by promoting flexible pedagogical approaches and maximising student connectedness. Multimodal learning analytics and indoor positioning technologies may hold promises to support teachers in complex learning spaces by making salient aspects of their spatial pedagogy visible for provoking reflection. This paper explores how granular x-y positioning data can be modelled into socio-spatial metrics that can contain insights about teachers' spatial pedagogy across various learning designs. A total of approximately 172.63 million position data points were collected during 101 classes over eight weeks. The results illustrate how indoor positioning analytics can help generate a deeper understanding of how teachers use their learning spaces, such as their 1) teaching responsibilities; 2) proactive or passive interactions with students; and 3) supervisory, interactional, collaborative, and authoritative teaching approaches. Implications of the current findings to future learning analytics research and educational practices were also discussed.