Leveraging artificial intelligence to assess the impact of COVID-19 on the teacher-student relationship in higher education.
Journal:
PloS one
PMID:
40127089
Abstract
The teacher-student relationship has far-reaching implications for educational outcomes at the tertiary level. Teachers contribute to students' success in various ways, including academic support, career counseling, personal mentoring, etc., that help them succeed academically and professionally. COVID-19 disrupted teacher-student interaction and hindered the flow of teacher's support to students. The damage caused by the pandemic to the higher education sector has mostly recovered. However, the trusting relationship between teacher and student is yet to get back to a pre-pandemic stage. Using stratified sampling technique, we collected nationally representative data from university students in Bangladesh and examined the relationship between COVID-19 and various aspects of the teacher-student relationship. We also explored the association between aspects of the teacher-student relationship and academic outcomes. In our sample, 28% of respondents are from STEM, and 72% are from non-STEM academic disciplines. We employed a subset of Artificial Intelligence (unsupervised machine learning) algorithms K-Modes clustering and Non-negative matrix factorization to cluster the data according to its internal structure. We created a new analysis technique called Absolute Rate of Fluctuation (ARF) to identify the fluctuations between the variables. ARF can track the fluctuations in any relationship induced by undesirable events such as the COVID-19 outbreak. We observed a deterioration in the interaction between teachers and students during COVID-19. However, the class conduction, exam taking, and assessment system were the most affected areas compared to personal interaction, catering support to students, and collaborative research activities.