The Cube³ lab is dedicated to improving the educational experience of undergraduate biomedical engineering (BME) students in and outside the classroom. The community-building branch of the lab focuses on the out-of-class experiences; the SoTL (Scholarship of Learning and Teaching) branch address teaching pedagogy in BME.

If you are a student reading this page, our community-building branch is our main research direction and is the area that needs the most help.

Community-Building Research

We have two active projects in community-building research.

Improving students' sense of belonging through informal student-faculty interactions

Formal (lectures, office hours) interactions between students and faculty may be intimidating for undergraduate students. We aim to facilitate student-faculty interactions outside of the classroom, in a more informal format, to break the communication barrier. We are interested in whether these informal interactions will foster a stronger sense of community in students. In the first phase of this research, we paired undergraduate students with faculties for a casual lunch. The lunches have been proven to be significantly beneficial to students' senses of belonging, clarity of career goals, and intention to graduate.

We are currently in Phase II of this study for characterizing whether the increases in sense of belonging have been equitable across multiple traditionally underrepresented groups in engineering. We also aim to gain qualitative insights from the students and faculties that we interviewed and refine our survey approach for larger-scale rollout.

Phase I of this work was supported by BME Health, Equity, and Wellness Committee and the Provost's Undergraduate Fellowship. Current phase (Phase II) is supported by the Academic Senate Large Grant.

Building a learning community for women in engineering

An important tone-setter for most college students is the first year or two years of their undergraduate experience. Negative experiences, such as the lack of community, during this vulnerable period may adversely affect women's sense of belonging in engineering and lead to potential dropout. Nationally, women have dropped out of engineering at a higher rate than men, despite having equal or superior math and physics scores coming out of high school. This learning community was formed to address this issue, better prepare the students for the academic environment you are about to enter, and possibly gain additional allies to tackle this problem.

We have started the Phase II of the study by integrating our first-year seminar experience with mentoring experiences to promote retention and persistence.

This work is currently supported by First-Year Seminar funds and the Provost's Undergraduate Fellowship.

Scholarship of Learning and Teaching Research

We have two active projects in SoTL research.

Problem-based learning in bioinstrumentation lectures

In our redesigned bioinstrumentation lectures, students are encouraged to participate real-life problem solving. We use a bedside electromyography (EMG)-based communication device as the theme of the redesigned bioinstrumentation course. In the course, students are taught relevant knowledge in sensors, conditioning, and digitization to recreate the themed device as a real-life design exercise, in addition to the lab modules they complete.

Collaborator: Dr. David Lin, Washington State University.

Developing Equity-Focused Module in Machine Learning Course

We aim to create a version of biomedical engineering-specific machine learning course that have a module in data and societal bias. The module contains a lecture in how training machine learning systems using data with bias, which could have multiple sources including intentional, non-intentional, societal, and representation biases, could result in a biased machine learning system that does harm. A hands-on module was developed along with the lecture to demonstrate gender bias in text processing systems.

Technical Research

We have one active project in machine learning research.

Better Detection of Atrial Fibrillation with Generated Electrocardiogram Signals

Training machine learning-based medical devices is difficult. The prevalence of most medical conditions is low, which results in class imbalance collected in real-life data sets. We aim to use generative neural networks to generate atrial fibrillation-like electrocardiogram (ECG) signals to improve the accuracy of detection in deep learning-based atrial fibrillation detectors.

Our AFE-GAN successfully generates ECG signals with atrial fibrillation characteristics that can trick existing atrial fibrillation detectors. Our next stage would be to demonstrate improvements to existing ECG detection systems.