Human vs. machine learning: enhancing student metacognion via the active learning of Physiology

Abstract

While machines outperform human associative memory, education should emphasise metacognition - including emotional intelligence. Physiology learning provides ample opportunity for cognitive development. Indeed, an updated mechanistic understanding of human body function and its dynamic regulation requires metacognition (awareness and understanding of thought processes, to self-assess, monitor and improve by self-regulation). This study implemented an assessment-based strategy to foster metacognition while students learn Physiology. Pharmaceutical Sciences students self-assessed their interest in Physiology and study strategies (Survey-1); monitored their skills in weekly formative tests (ForTes - including a self-prediction of score and percentile); and received counselling on active study techniques. Students reported their study adaptations in Survey-2; and were rewarded for accurate predictions of performance in Exams. Anonymised data from ForTes and surveys were analysed with R language to compute statistics and perform machine learning (ML). Unsupervised ML of Survey-1 identified 3 student clusters with different levels of motivation and study, which associated with different Exam scores. A supervised ML model trained with ForTes predicted Exam failures with over 90% accuracy, and the actual Exam scores (0-20) with error under 2 points. Analysis of ForTes over time evidenced significant improvements in students' metacognitive ability to accurately self-assess, despite being challenged with different questions about different Physiological systems over time. Moreover, the frequent report of study adaptations in Survey-2 supports widespread metacognitive regulation. This study provides evidence-based support for enhanced student metacognition and suggests that ML models can accurately predict student failure, enabling early counselling towards success. This study also emphasises the need for authentic assessment to promote Physiological skills.