Improvement in educational performance through wearable-based flow predictive models

Flow Theory has been used to study motivation in educational activities. However, few studies use physiological data to uncover unknown aspects of said data in any context, and isolated individuals are involved as well. In this paper, we present some of the results obtained from two control groups corresponding to two full primary education classrooms, as well as their teacher, using a quasi-experimental design. They participated in two training activities with different instructional designs and three different STEAM subjects: graphic design, video game design using Roblox Studio, and educational robotics. In this sense, the heart rate, its variability, data from accelerometers, and the educational activities carried out by the teacher have been automatically recorded for each participant at every second. To achieve this, we used smartwatches connected to Polar H10 sensors as well as our own apps. At the end of each session, everyone answered the Flow FKS and EduFlow prevalence questionnaires, and the teacher kept a class journal. Through this, we aim to understand whether the Flow Theory models derived from the FKS and EduFlow scales are valid from a physiological standpoint, as well as to develop classification and predictive models based on artificial intelligence that will allow for educational performance improvement of students in future research.