STEM Persistence (Current)
A well-documented disparity (eg gender and race gap) in Science, Technology, Engineering, and Mathematics ( fields, has received research attention for several decades In particular, the mechanisms of learning have been the focus of scholars in philosophy, psychology, education, and computer science in order to identify predictors of these disparities however, this growing body of research has also raised more questions than answers Existing research has investigated various cognitive, emotional, motivational, and institutional components of learning of the underrepresented students, but fail to account for context and dynamics of the students’ learning behavior To understand this, we propose to examine the research question How do the emotional lives of underrepresented STEM students impact their learning behavior, in particular, their STEM persistence? Specifically, why do some of them succeed in STEM persistence, while others do not?
In collaboration with CWIT, DoIT, Psychology and Honors College at UMBC
STEM Career Prediction
The increasing capabilities of intelligent tutoring systems (ITS) that collect student interaction data while learning mathematics at the middle school level have enabled researchers in educational data mining (EDM) to develop models that predict student career choice. Current research focuses on feature selection techniques that provide essential features in predicting the target variable. However, the factors that affect the prediction performance of an algorithm at a sample level could be studied in depth as they influence the overall performance of an algorithm. In this study, we analyze the influence of various attributes collected by the ASSISTments online learning platform on the performance of machine learning algorithms in predicting student career fields. This proposed study facilitates researches in the field of EDM with factors that influence the development of efficient models in predicting STEM and Non-STEM careers.
Affect Change Model
Affect detection in educational systems has a promising future to help develop intervention strategies for improving student engagement. To improve the scalability, sensor-free affect detection that assesses students’ affective states solely based on the interaction data between students and computer-based learning platforms has gained more and more attention. In this paper, we present our efforts to build our affect detectors to assess the affect changes instead of affect states. We developed an affect-change model to represent the transitions between the four affect states; boredom, frustration, confusion and engagement concentration.
ACL Rehabilitation Tracking (Current)
Anterior Cruciate Ligament Rupture (ACLR) is one of the most common injuries among athletes and recreational sports players. Annually, 68.6 ACL injuries occurred per 100,000 general people in the United States in 2010, and this number is even higher in athletes. Over time, ACLR can also lead to osteoarthritis, ACLR in the opposing knee, and even reinjuring the same knee. In this study, we try to work on a research question “Can a person’s risk of ACLR (re)injury and game readiness, be predicted by analyzing their GAIT cycle using computational models utilizing various machine learning algorithms, sensors, and classification types?”
In Collaboration with Link Lab and Exercise and Sports Injury Lab (EASIL) at the University of Virginia, Charlottesville
Neurophysiological variations in food decision making (Current)
Epidemiological studies have revealed a strong association between disordered eating behavior and obesity-related health consequences. The development and onset of dysfunctional eating behavior and eating disorders begin in adolescence and young adulthood (DSM-V). Moreover, weight gain occurs during the transition to college, where young adults make more independent food and diet-related decisions. Examining food and eating-related processes during this developmental period among college students will help develop an effective treatment for individuals with subthreshold eating symptomatology or eating disorders. This pilot study aimed to achieve such a long-term goal by designing and developing experimental environments, protocols, and methods to examine neurophysiological variation in food decision-making processes within a Virtual Reality (VR) and a Real Life (RL) buffet environment. We explored two research questions:
What are the neurophysiological correlates of food decision-making behaviors within the VR and the RL buffet setting?
How consistent are individuals’ food decision-making processes in the VR setting and the RL buffet setting?
In Collaboration with CCAD Lab (Psychology), Imaging Research Center (Visual Arts) at UMBC and Epidemiology at Dartmouth