Physiological Impact of Test Anxiety on Student’s Academic Performance Using Convolution Neural Network

Abstract

Using a Convolutional Neural Network (CNN), this research studies the connection between physiological reactions and students' performance when taking examinations. They used a full PhysioNet dataset that included things like accelerometer data, skin surface temperatures, interbeat interval, heart rate, blood volume blood pressure, and electrodermal activity (EDA). Ten students had their data gathered from three separate assessments: the first midterm, the second midterm, and the final. By analyzing these physiological signals, aimed to identify patterns and correlations that indicate how students respond to exam stress and cognitive load.Our CNN model achieved an accuracy of over 98%, surpassing the performance reported in similar studies [21][24], which utilized deep neural networks for emotional intelligence and educational data mining, respectively, achieving slightly lower accuracy. This advancement highlights the novelty of our approach to accurately predicting academic performance based on physiological data.The findings reveal that students experience heightened physiological responses, such as increased heart rate and skin surface temperature, during exams, indicating elevated stress levels. Notably, there is a significant correlation between these responses and students' grades, suggesting stress levels can negatively impact academic performance.This research advances our understanding of the physiological underpinnings of test anxiety and its effects on academic outcomes. By leveraging physiological metrics, this study offers the potential for developing predictive models and interventions to support student well-being and academic success. The insights gained could lead to more comprehensive assessment practices in education, ultimately aiding in the creation of supportive learning environments that prioritize students' mental and physical health.